Immunohistochemistry is a valuable tool in routine breast pathology, used for both diagnostic and prognostic parameters. The diagnostic immunomarkers are the scope of this review. Most breast lesions can be diagnosed on routine hematoxylin-eosin sections; however, in several scenarios, such as morphologically equivocal cases or metastatic tumors of unknown primary, the appropriate application of immunohistochemistry adds true value in reaching an accurate diagnosis.
To evaluate the diagnostic utility of the most commonly studied immunomarkers in the field of breast pathology by review of the literature, using the database of indexed articles in PubMed (US National Library of Medicine, Bethesda, Maryland) from 1976 to 2013.
Literature review, and author's research data and personal practice experience.
The appropriate use of immunohistochemistry by applying a panel of immunomarkers and using a standardized technical and interpretational method will complement the morphologic assessment and aid in the accurate classification of difficult breast lesions and the identification of metastasis from a breast primary.
There are well-defined morphologic criteria for classifying most breast lesions; however, not infrequently, we encounter cases with equivocal morphologic features requiring ancillary studies, especially immunohistochemistry, to reach an accurate diagnosis. The common scenarios that require immunohistochemistry and the application of immunomarkers are summarized in Table 1. In addition, several immunomarkers are useful when working on tumors from an unknown primary to aid in the identification of breast origin. The application of those immunomarkers and their pitfalls are discussed below.
THE EVALUATION OF INVASION
The World Health Organization's Pathology and Genetics of Tumours of the Breast and Female Genital Organs states “invasive breast carcinoma is a group of malignant epithelial tumours characterized by invasion of adjacent tissues and a marked tendency to metastasize to distant sites.”1(p13) Histologically, the hallmark of invasion is the lack of myoepithelial cells (MECs),2 which functionally are a hybrid of both smooth muscle (“myo,” with contractile property) and epithelial cells (with cadherin-mediated cell-cell junctions), and immunohistochemically express filamentous smooth muscle actin (SMA) and smooth muscle myosin as well as intermediate filaments (the epithelial keratins).3 However, the presence or absence of MECs is not always easily appreciated on routine hematoxylin-eosin sections. Throughout the years, researchers discovered several immunomarkers that target different proteins in MECs to provide an objective measure of evaluating them when encountering challenging cases, to differentiate carcinoma in situ (CIS) (ductal [DCIS] or lobular [LCIS]) or sclerosing adenosis from invasive breast carcinoma (CA) and benign or atypical papillary lesions from papillary CAs. An example of the differential staining pattern of sclerosing adenosis versus tubular CA is illustrated in Figure 1, a through d, and Figure 2, a and b. The commonly used MEC markers and their immunostaining patterns and pitfalls4 are summarized in Table 2 and are discussed below.
p63, a p53 homologue, detects MECs in breast with a dotlike nuclear staining pattern as illustrated in Figure 1, b. p63 is considered one of the best MEC markers with clean background, for example, no cross-reaction with stromal myofibroblasts or vascular smooth muscle cells; however, it may show focal gaps (discontinuous pattern) around noninvasive epithelial nests (especially with CIS)5–21 and may also label adenoid cystic CA and metaplastic CA of the squamous component in a diffuse fashion, as illustrated in Figures 3, a through d. In papillary lesions, especially papillary CA, p63 may show focal patchy reactivity in tumor cells in up to 33.3% of cases.17,18 Because of its dotlike, noncontinuous nuclear-staining pattern, the evaluation may not be very easy in lesions such as sclerosing adenosis and papillary lesions. A combination of p63 and other MEC markers, such as smooth muscle myosin heavy chain (SMMHC) and calponin, is often suggested.
Smooth muscle myosin heavy chain is a 200 kDa, unique structural component of myosin, widely used to identify epithelial cells in breast pathology practice. Smooth muscle myosin heavy chain shows a continuous cytoplasmic staining pattern in normal and benign lobules or ducts and may show gaps in CIS. Compared with p63, SMMHC may have a slightly higher sensitivity; however, cross-reactions with stromal myofibroblasts and vascular smooth muscle cells have been reported,5–8,10,15–18,22 although in lower frequency, when compared with calponin (8% versus 76%).
Calponin, a 34-kDa polypeptide, is expressed extensively in smooth muscle tissue and is involved in the contraction mechanism. Calponin is one of the MEC markers demonstrating a continuous cytoplasmic linear staining pattern in normal or benign breast tissue, with a focal discontinuous pattern in a few CISs.5–10,14–16,18,20,21 A high frequency of cross-reactivity with stromal myofibroblasts and vascular smooth muscle cells was reported20 ; in addition, cross-reactivity with tumor epithelial cells was also noted in 18% of cases.
CD10—cluster of differentiation 10 or common acute lymphoblastic leukemia antigen—is an integral membrane glycoprotein expressed in MECs with a cytoplasmic staining pattern.23 Studies5–8,14,18–21,24 showed that CD10 is a relatively sensitive MEC marker but has cross-reactivity to myofibroblasts and nonspecific reactivity to epithelial cells. Tse et al19 reported an overall sensitivity of 91%, cross-reactivity to myofibroblasts in 20% of cases, and nonspecific epithelial reactivity in 28% of cases. CD10 was reported as showing no reactivity to vascular smooth muscle cells.
Maspin, the product of a candidate tumor suppressor gene related to the serpin family of protease inhibitors, is intensely expressed in MECs of mammary gland.25 Maspin is a sensitive MEC marker with both a nuclear and cytoplasmic staining pattern and clean background without cross-reactivity to stromal myofibroblasts or vascular smooth muscle cells.7,26–28 However, a proportion of invasive breast CA and DCIS results are positive for maspin.29–31 Our unpublished data on the immunohistochemical study of maspin expression in breast CAs and normal tissues revealed the following findings: (1) 100% sensitivity for MECs; (2) no cross-reactivity with myofibroblasts or vascular smooth muscle cells, as also reported by others; and (3) 29% (75 of 259) of the invasive breast CAs expressed maspin in a partial to diffuse pattern. The staining patterns of maspin are illustrated in Figure 4, a through d.
Other MEC markers, such as SMA, p-cadherin, Wilms tumor 1 (WT1), S100, and high–molecular-weight cytokeratin (HMWCK) or basal-type cytokeratin (CK) (CK5/6, CK14, CK17, CK903) are less commonly used because of marked cross-reactivity with myofibroblasts and vascular SMA,4–9,19,20 frequent reactivity in tumor cells (HMWCK or basal-type CK, S100, p-cadherin),4–8,19,32–34 or low sensitivity (WT1 and S100).33 Podoplanin (D2-40), a marker labeling lymphatic endothelium, was observed to be variably expressed in breast MECs surrounding benign ducts and lobules as well as in CISs.35 However, its utility as a MEC marker has not been evaluated. p75NTR, the low affinity neutrophin receptor, was reported to be consistently positive for MECs in mammary gland. Its immunoreactivity was comparable to p63 and SMMHC.36
DIFFERENTIAL DIAGNOSIS OF LOBULAR NEOPLASIA VERSUS DUCTAL NEOPLASIA
The World Health Organization's Pathology and Genetics of Tumours of the Breast and Female Genital Organs1 states invasive ductal CA, not otherwise specified, “is a heterogeneous group of tumours that fail to exhibit sufficient characteristics to achieve classification as a specific histological type,”(p19) whereas invasive lobular CA “usually associated with lobular carcinoma in situ is composed of non-cohesive cells individually dispersed or arranged in single-file linear pattern in a fibrous stroma.”(p23) Histologically, there are characteristic features to distinguish ductal from lobular CAs, such as cell-to-cell cohesion, lumen formation, and a defined tumor cell border, which favors ductal CA, and monotonous, small, discohesive cells, individually dispersed through, or arranged in a single-file pattern in, the fibrous connective tissue, which favors lobular CA. Most ductal or lobular CAs can be diagnosed on routine hematoxylin-eosin sections. However, we encountered cases with equivocal histomorphologic features, such as solid-patterned CAs, either in-situ or invasive; under those circumstances, the distinction between ductal and lobular carcinoma based solely on morphology is challenging. Immunohistochemistry is a helpful adjunct to the accurate classification of those challenging cases. The differential immunophenotypes4 are summarized in Table 3 and discussed below.
E-cadherin, the product of the CDH1 gene (16q22.1), is a transmembrane cell adhesion molecule comprising a cytoplasmic domain, which interacts through intracellular catenins with the actin-based cytoskeleton, and an extracellular domain, which is involved in homotypic cell-to-cell adhesion.37–39 This cadherin-catenin system is important in the organization and integrity of most epithelial tissues. Loss of E-cadherin function and/or a dysfunctional cadherin-catenin complex are common findings in lobular neoplasia, which, at a molecular genetic level, are caused by CDH1 gene aberrations with several mechanisms, such as somatic mutation of CDH1 with loss of heterozygosity on chromosome arm 16q and epigenetic changes, homozygous deletion of the CDH1 gene, and CDH1 gene promoter hypermethylation.40–42
Immunohistochemically, the molecular events correlate with a complete loss of expression of E-cadherin or aberrant localization of E-cadherin protein (cytoplasmic, as apical or perinuclear).43 Morphologically, loss or abnormal function of the E-cadherin-catenin complex determines the phenotypic features for lobular neoplasia, such as dyshesive tumor cells, diffuse infiltrative growth pattern, and a distinct metastatic pattern.44 Numerous studies have investigated the expression of E-cadherin protein in lobular and ductal CAs and have revealed a complete loss of E-cadherin expression in most lobular CAs, in contrast with the diffuse membranous staining for E-cadherin in most ductal CAs. However, aberrant E-cadherin expression was identified in occasional lobular CA cases, which was reported in the range of 2% to 16%.45–49 Harigopal et al50 reported aberrant E-cadherin staining patterns in 5 invasive mammary CAs. In their study, the aberrant E-cadherin expression was defined as an E-cadherin immunophenotype that did not correspond to the apparent histologic classification of the lesion. Two of the 5 primary invasive CA cases (40%) contained apparent ductal components that failed to show E-cadherin membranous staining. One of the 5 primary invasive CAs (20%) and one metastatic CA (20%) were apparently lobular CAs showing strong membranous staining for E-cadherin. Several authors suggested that the expression of E-cadherin in tumors showing characteristic features of lobular CA should not preclude the diagnosis of lobular CA.
p120 catenin, which was initially described as a prominent substrate of the Src oncoprotein,51 is encoded on band 11q11 and belongs to the catenin family.52 A major role of catenins (including α-catenin, β-catenin, plakoglobin, and p120 catenin) is to anchor the E-cadherin complex to the actin cytoskeleton. The α-catenin and β-catenin are complexed with the carboxy-terminal cytoplasmic tail of E-cadherin, and the p120 catenin is anchored to E-cadherin in a juxtamembranous site.53 p120 catenin is mostly bound to E-cadherin in the membranes, with a minor cytoplasmic pool of p120 catenin. Immunohistochemically, p120 catenin is detected in the cell membranes of a variety of epithelial and nonepithelial tissues.54 Shibata et al55 studied 60 cases of lobular CA; they found 55 cases (92%) lacked E-cadherin expression and also lost membranous staining for p120 catenin, showing instead a cytoplasmic staining pattern. In contrast, the remaining 5 cases (8%), which showed intact E-cadherin expression, were p120 catenin positive in a membranous pattern. That study suggests p120 catenin and E-cadherin are colocated in the lateral membranes of normal mammary epithelial cells, and p120 catenin is localized in the cytoplasm of E-cadherin-deficient lobular cancer cells. Several studies evaluated p120 catenin expression in non-neoplastic breast tissue, lobular neoplasia, and ductal CA and have observed intense, linear, membranous immunostaining in normal tissue and ductal carcinoma, whereas there was intense cytoplasmic immunostaining in lobular neoplasia (in situ and invasive).56–59 Our previous study of p120 catenin expression in 44 invasive lobular CAs (50%), 18 LCISs (20%), and 26 low-grade ductal CAs (30%) revealed invasive lobular CAs that showed strong and diffuse cytoplasmic staining in 77% (34 of 44) of the cases and moderate to weak cytoplasmic and membranous staining in 23% (10 of 44) of the cases; LCISs showed strong cytoplasmic staining in 89% (16 of 18) of the cases. In contrast, all invasive ductal CA cases were positive for p120 catenin, with a strong and diffuse membranous staining (>3+) in 22 of 26 cases (85%). The remaining 4 invasive ductal CA cases (15%) showed focal (1+ or 2+) and weak membranous staining. Strong cytoplasmic positivity was not observed in invasive ductal CA. Nonneoplastic ductal epithelium was positive for p120 catenin with predominantly membranous staining pattern.60 Examples of the p120 catenin staining pattern in ductal and lobular CAs are illustrated in Figure 5, a through b. However, in the Rakha et al49 study, 16% (38 of 239) of the invasive lobular CAs showed aberrant E-cadherin expression (membranous staining for E-cadherin). Further studies for catenin expression were conducted in E-cadherin–positive cases, revealing that 54% (13 of 24) of E-cadherin–positive invasive lobular CAs showed moderate to strong membranous immunostain for p120 catenin in addition to moderate to strong cytoplasmic staining, which was not observed in ductal CA. Brandt et al61 reported 10% of ductal CAs showed weak membranous staining for p120 catenin, which could, therefore, be misinterpreted as a lobular phenotype. Those data suggest p120 catenin should be used in conjunction with E-cadherin and interpreted in accordance with histomorphology.
β-Catenin, a transcription cofactor with T-cell factor/lymphoid enhancer factor in the Wnt pathway, also binds tightly to the cytoplasmic domain of type I cadherins (such as E-cadherin), linking cadherins through α-catenin to the actin cytoskeleton.62 In breast CAs, differential staining patterns for β-catenin showed that most lobular CAs are negative for β-catenin or have cytoplasmic granular staining patterns, whereas ductal CAs show membranous staining with or without cytoplasmic staining.41,56,59,63 However, in invasive lobular CAs with aberrant E-cadherin expression, more than one-half of the cases also expressed β-catenin (membranous staining pattern).48,49
Cytokeratins are intermediate filaments that form a complex cytoskeleton network within cells and interact with the cytoskeleton of neighboring cells. Studies observed differential immunostaining patterns for CK903 (clone 34βE12) and CK8 (clone CAM 5.2) in lobular neoplasia and ductal neoplasms.38,64,65 Bratthauer et al64 reported that CK903 demonstrated a perinuclear cytoplasmic staining pattern in 100% (40 of 40) of the classic lobular intraepithelial neoplasias (atypical lobular hyperplasia and LCIS) and, in contrast, was negative in all 20 cases (100%) of classic ductal intraepithelial neoplasias (atypical ductal hyperplasia [ADH] and DCIS). Lehr et al65 reported that CK8 demonstrated a cytoplasmic peripheral-predominant staining pattern with neighboring cells molding together in ductal CAs (33 of 33 cases; 100%) and a ring-like perinuclear immunostaining pattern creating a “bag of marbles” appearance with neighboring cells in lobular CAs (15 of 15 cases; 100%). Examples of these immunostaining patterns are illustrated in Figure 6, a through d.
DIFFERENTIAL DIAGNOSIS OF USUAL DUCTAL HYPERPLASIA VERSUS ADH/DCIS
Widespread use of mammography has resulted in increased detection of breast intraductal proliferative lesions comprising a spectrum of lesions, from benign (usual ductal hyperplasia [UDH]), borderline (ADH), to preinvasive (DCIS). However, interobserver reproducibility in the diagnosis of intraductal proliferative lesions is poor, giving rise to potential misclassifications in treatment protocols.66 In human mammary glands, the ductal and lobular units are composed of the luminal epithelium, which express CK7, CK8, CK18, and CK19, and the myoepithelial (basal) layer containing CK5, CK7, CK14, and CK17.67 Several studies have investigated the expression of low–molecular-weight cytokeratins (CK8, CK18, and CK19) and HMWCKs or basal cytokeratins (CK903, CK5/6) in normal, benign, borderline, and malignant breast lesions. Such studies have found a differential pattern of expression of basal cytokeratins (HMWCK) in UDH versus ADH/DCIS, that is, high expression was found in UDH and no or focal weak expression was found in ADH/DCIS.68–74 By application of a breast marker cocktail (CK5, CK14, CK7, CK18, and p63), UDH exhibits a mosaic staining pattern, and ADH and DCIS show clonal proliferation of luminal epithelial cells with negative HMWCK (basal cytokeratin) staining. The immunophenotypes reflect growing evidence that UDH is a hyperplastic process, whereas ADH and DCIS are neoplastic, with a clonal proliferation of luminal epithelial cells.75–78 Estrogen receptor (ER) expression is scattered in UDH and diffuse in ADH/DCIS, reflecting its clonal proliferation. The differential staining patterns4 are summarized in Table 4.
THE EVALUATION OF PAPILLARY LESIONS
Papillary lesions of the breast include a spectrum of benign (intraductal papilloma), atypical (ADH involving intraductal papilloma), and in situ (papillary DCIS and DCIS involving intraductal papilloma) to invasive lesions (solid papillary CA), which often pose diagnostic challenges to the practicing pathologist. In addition, intracystic papillary CA, for which the term encapsulated papillary carcinoma has been proposed, is considered an in situ lesion; however, the lack of MECs at periphery of this lesion may suggest that it is a part of the progression, an intermediate between in situ and invasive CA.79 Because of the diagnostic challenges in the distinction of benign, atypical, and malignant papillary lesions based on hematoxylin-eosin morphology, many investigators explored the diagnostic utility of immunomarkers, mainly MEC markers and HMWCKs, in the accurate classification of papillary lesions. The practical approach and application of immunohistochemistry in papillary lesions are illustrated in Figure 7 and discussed below.
Immunohistochemical evaluation of MECs at periphery of lesional epithelium has been used as a means to differentiate invasive from noninvasive lesions.5,6 In papillary lesions, several studies16–21,72,80 investigated MEC expression, reported in 100% of benign intraductal papillomas with or without ductal hyperplasia and 0% or focally, weakly attenuated in intracystic (encapsulated) papillary CAs. In the assessment of atypical papillary lesions (ADH/DCIS involving benign intraductal papilloma or papillary DCIS), immunohistochemical evaluation of the expression of HMWCK is a valuable adjunct, in which benign papillary lesions exhibit strong, mosaic reactivity throughout the lesion in 88% to 100% of cases, and atypical papillary lesions are nonreactive, indicating a clonal proliferation of luminal epithelial cells in 80% to 100% of cases.19,20,68,69,81–83 Hormonal receptors, such as ER, may have additional value; in general, benign lesions show only scattered staining, whereas atypical papillary lesions are usually diffusely positive.77,78,84,85 In addition, neuroendocrine markers, such as synaptophysin and chromogranin, have been reported to be positive in most solid papillary CAs and negative in benign and atypical papillary lesions.20,86–88
THE EVALUATION OF SPINDLE CELL LESIONS
Spindle cell lesions of the breast are rare but often pose diagnostic challenges, especially in limited needle core biopsy specimens. The most common monophasic spindle cell lesions include fibromatosis, myofibroblastoma, and inflammatory myofibroblastic tumors (pseudotumor); metaplastic (sarcomatoid) CA; and the very rare primary breast sarcoma. Tumors with a biphasic pattern are mainly fibroepithelial tumors, including fibroadenoma and phyllodes tumor (PT), and biphasic metaplastic CA.89 Immunohistochemical studies are often employed in the workup of spindle cell lesions of breast, especially in the group with a monophasic pattern. The most important task is to exclude the potential diagnosis of metaplastic CA. The practical approach and application of immunohistochemistry in the workup of spindle cell lesions are illustrated in Figure 8. The common entities are discussed below.
Metaplastic CA of the breast is a heterogeneous group of neoplasms generally characterized by an intimate admixture of adenocarcinoma with dominant areas of spindle cell, squamous cell, and/or mesenchymal cell differentiation; the metaplastic spindle cell and squamous cell CAs may present in a pure form without any admixture with a recognizable adenocarcinoma.1 Studies show that cytokeratin immunoreactivity may be focal; therefore, a broad panel of low–molecular-weight cytokeratins, HMWCKs, and pancytokeratins should be applied, in addition to MEC markers, such as p63, calponin, S100 protein, and SMA. Pancytokeratins (MNF116) and HMWCKs, also known as basal cytokeratins, such as CK903, CK5/6, and CK14, were among the most sensitive markers to detect cytokeratin expression in this setting.8,89–93 CK7, CAM 5.2 and AE1/3 are usually negative or only focally positive; p63 is often positive; CD34 is nonreactive and may be used in a panel to differentiate metaplastic CA from myofibroblastoma and PT.93–100 Sex-determining region Y box 10 (SOX10) has been documented as a MEC marker in benign breast tissue and was reportedly positive in 46% (6 of 13) of MCAs of the spindle cell component. In contrast, none of the 44 cases (0%) of the fibroepithelial neoplasms, including 10 FAs and 34 PTs, were reactive.101
Fibroepithelial tumors of the breast, including fibroadenoma and PT, are a heterogeneous group of biphasic lesions combining an epithelial component and a quantitatively predominant mesenchymal component (also called a stromal component).1 Depending on the composition of the tumor, in cases with stromal overgrowth, especially in limited biopsy material, the epithelial component may not be easily identified. The diagnosis of metaplastic carcinoma should always be considered and be excluded by applying a broad spectrum panel of cytokeratins and MEC markers. In addition, the histologic differentiation between fibroadenoma and PT and the grading of PT are challenging on limited biopsy tissue. Investigators explored the expression of several immunomarkers in fibroepithelial tumors of the breast, attempting to identify the differential staining pattern. Among those, the Ki-67 proliferative index has been reported to be increased in borderline and especially in malignant PT and can be useful in the classification of fibroepithelial tumors.102–106 The World Health Organization classification of PT1 requires greater than 10 mitoses per 10 high-power fields for malignant PT; however, no numeric cutoff has been established to define benign and borderline PTs, which were described as “few if any” and “moderate,” respectively. Some authors2 classify less than 2 mitoses per 10 high-power fields as benign PT, 2 to 5 mitoses per 10 high-power fields as borderline PT, and more than 5 mitoses per 10 high-power fields as malignant PT. Several immunomarkers showed increased expression in the stroma of malignant PT compared with that in benign tumors, including p53,8,103,105,107–112 CD117,8,103,105,107–112 epithelial growth factor receptor (EGFR),113 and CD10.114 However, no significant difference can distinguish between borderline and malignant PT. Esposito et al103 reported epithelial endothelin 1 negativity was associated with malignant PT. CD34 and B-cell chronic lymphocytic leukemia (CLL)/lymphoma 2 (Bcl2) are reported to be expressed in the stroma of most fibroepithelial tumors115–117 but not in metaplastic CA. Nuclear localization of β-catenin is a common feature of fibromatoses. It is also frequently identified in the stroma of PT but rarely noted in 23% (12 of 52) of MCAs, although it is often focally expressed in those positive cases.118,119 Most recently, Yang et al105 observed that insulin-like growth factor II messenger RNA binding protein-3 (IMP3) was preferentially expressed in all malignant PTs but not in borderline or benign tumors or benign surrounding breast tissues. Despite the research efforts, histomorphology remains the gold standard for the diagnosis of fibroepithelial tumors.120
MARKERS THAT AID IN THE IDENTIFICATION OF BREAST PRIMARY
Estrogen receptor, gross cystic disease fluid protein 15 (GCDFP-15), and mammaglobin-A (MGB) are the most commonly used breast-specific immunomarkers in the workup of tumors of an unknown primary.121–125 In metastatic breast CAs, the ER+ rate is about 50%.126–133 Both GCDFP-15 and MGB suffer low sensitivities, reported in the literature in the ranges of 35% to 55% and 65% to 70%, respectively. Our data on TMAs of 250 cases of invasive breast CAs, including ductal, lobular, and other special types, was even lower: 30% for GCDFP-15 and 50% for MGB, which is similar to the reports by Bhargava et al122 who found GCDFP-15 expression in 23.1% and MGB expression in 55.4% of breast CAs, and by Lewis et al125 who reported GCDFP-15 labeling of 37% and MGB labeling of 54% in breast CAs. Given the frequent absence of expression of currently available breast-specific immunomarkers (such as ER, GCDFP-15, and MGB) in metastatic breast CAs, studies to discover newer breast-specific immunomarkers were undertaken. NY-BR-1 (also named Ankyrin repeat domain 30A) and GATA binding protein 3 (GATA3) are among the most promising immunomarkers.
NY-BR-1, a differentiation antigen of the mammary gland, was first described in 2001 by Jager et al134 using SEREX (a serologic analysis of recombinant tumor complementary deoxyribonucleic acid expression libraries) in a patient with breast cancer. They reported that NY-BR-1 messenger ribonucleic acid (mRNA) expression was restricted to normal breast and testis tissues (although at a much lower level) and to 84% (21 of 25) of breast CAs; whereas a variety of other normal tissues and most other tumor tissues showed a lack of expression. Since then, studies were undertaken to explore NY-BR-1 protein expression in normal and tumor tissues.135–142 The immunostaining pattern was described as mainly cytoplasmic with focal nuclear staining. In invasive breast CAs, NY-BR-1 expression was reported in 46.6% to 70% of cases, showing a strong association with ER+ and lower-grade CAs.135–138,141,142 In ER− breast CAs, its expression was reported as much lower, 18% by Woodard et al141 and 28.4% by Balafoutas et al.142 In normal tissues, NY-BR-1 expression was noted exclusively in mammary gland epithelia135–138,141,142 ; reactivities in skin adnexal structures140 as well as focal, weak staining in Barrett esophagus and benign mucosal glands of the stomach were also noted.141 In other tumors studied, NY-BR-1 expression was noted in 27% (3 of 11) of sweat gland carcinomas,138 75% (18 of 24) of mammary Paget disease, 81% (21 of 26) of extramammary Paget disease,140 5.6% (8 of 142) of müllerian carcinomas, and 7% (1 of 15) of pancreatic tumors.141 The current data in the literature suggest the potential diagnostic utility of NY-BR-1 as a breast-specific immunomarker; however, a larger series of studies is necessary to validate those findings.
GATA3, also known as GATA binding protein 3, is a member of the group of 6 zinc-finger transcription factors. It regulates the specification and differentiation of tissues, such as breast, kidney, nervous system, parathyroid gland, hair follicle, placenta trophoblasts, thymocytes, and T cells.143–154 In normal tissues,155 GATA3 expression (labeling nucleus) was noted in 50% (5 of 10) of mammary gland tissues in the luminal epithelial cells in a patchy fashion, 100% (10 of 10) of the normal urothelium in a diffuse pattern, and 100% (20 of 20) of the normal renal tissues in the distal renal tubules, whereas the proximal renal tubules, glomeruli, and interstitium in normal renal tissue; the MECs in normal mammary gland tissue; and 270 other varieties of normal tissues were nonreactive. Gene expression profiling studies have demonstrated that GATA3 is highly expressed in a subset of human breast tumors156–158 and that GATA3 expression in breast tumors highly correlates with expression of the estrogen receptor α gene/protein (ESR1)159 and is seen in tumors of the “luminal A” subtype.160 In 2007, Higgins et al161 analyzed expression patterns in prostate and bladder cancer tissues using complementary DNA microarrays. They found that the GATA3 gene showed high-level expression in urothelial carcinomas. Further immunohistochemical evaluation of GATA3 expression in 1027 varieties of CAs was undertaken. GATA3 expression was reported in 100% (4 of 4) of the breast ductal CAs in addition to 67% (206 of 308) of urothelial CAs but none of the other tumors tested. To substantiate those findings, our previous study155 investigated GATA3 expression in 1110 malignant neoplasms and 310 normal tissues. We found GATA3 expression in 94% (138 of 147) of the breast CAs, 86% (62 of 72) of the urothelial CAs, 2% (2 of 96) of the endometrial CAs but no expression in other tumors tested (n = 795). An additional study of GATA3 expression in 96 ER− breast CAs,162,163 including 6 MCAs, yielded a positive rate of 69% (66 of 96), but only 16.7% (1 of 6) in the MCAs; in comparison, GCDFP-15 and MGB expressions were evaluated as well in that study, revealing a positive rate of 15% (14 of 96) and 35% (34 of 96), respectively. Representative photos are shown in Figure 9, a through d. In 2013, Cimino-Mathews et al164 studied 99 invasive breast CAs and 34 fibroepithelial neoplasms and found GATA3 expression in 67% (66 of 99) of the breast CAs and stromal GATA3 labeling in only 1 of 34 fibroepithelial neoplasms (3%); the positive rate was 43% in triple-negative tumors and 54% in MCAs. Wu et al165 reported GATA3 expression in 66.7% (40 of 60) of ER− breast CAs. However, 2 other studies reported GATA3 expression in ER− breast CAs: Yang et al166 at 5% and Albergaria et al167 at 16%. Overall, the current available data suggest that GATA3 is a sensitive and relatively specific immunomarker for breast and urothelial CAs; among the breast-specific markers available currently, GATA3 appears superior to others. However, more data are emerging, reporting GATA3 expression in 49% (81 of 164) of salivary gland tumors,168 95% (20 of 21) of pheochromocytomas,169 89% (31 of 35) of paragangliomas,169 96% (24 of 25) of benign Brenner tumors of the ovary,170 100% of parathyroid tumors,169 and 0% to 23% of squamous cell CAs of lung.155,171–173 In practice, when interpreting GATA3 results, caution should be exercised.
Immunohistochemistry has an important role in routine diagnostic breast pathology practice, especially in the evaluation of invasion, papillary lesions, and spindle cell lesions to exclude metaplastic CA. Among the breast-specific markers, GATA3 is superior to others and should be used in a panel when working on metastases.
The author has no relevant financial interest in the products or companies described in this article.