Immunohistochemistry is helpful in cases of ACC, not only to confirm the diagnosis, but also to differentiate it from its mimickers. A panel including C–KIT (CD117), P63, heavy chain myosin, and calponin is very helpful [6]. ACC is usually positive for C–KIT and P63, and negative for both heavy chain myosin and calponin. Collagenous spherulosis and cribriform DCIS are positive for all myoepithelial cell markers but negative for C–KIT, while invasive cribriform carcinoma will not express any of those markers. Although, typically negative for ER, PR, and HER-2, up to 12 % of mammary ACC have been reported to be ER+/PR+ [5]. Other immunohistochemical studies dedicated solely to mammary ACC are identified in the literature, but are all limited by the small number of cases studied. These studies show a low proliferation index manifested by Ki-67, lack of P53 expression [7], and low Topoisomerase IIα expression, while demonstrating an over-expression of EGFR in 65 % of ACC cases [8].

Microarray–based gene expression studies have included ACC in the basal-like category due to their triple negative phenotype and expression of basal cell markers. However, studies focusing only on ACC show some molecular differences that distinguish ACC from the crowd of TNBC. ACC consistently displays t(6;9)(q22–23; pp 23–24) translocation, which generates a fusion transcript involving the MYB and NFIB genes [9]. This translocation is considered the key oncogenic mechanism in the pathogenesis of ACC. EGFR gene amplification which has been shown in some basal-like breast cancers, has not been demonstrated in ACC [8]. Similarly, C-KIT expression characteristic of ACC does not reflect an underlying KIT mutation.

The diagnosis of metaplastic carcinoma in challenging cases lies in the identification of the epithelial origin of the tumor cells. This may require the use of a panel of low-and high-molecular weight cytokeratins since many metaplastic carcinomas show only focal positivity for CK or may even be negative to some. CK903(34betaE12) and P63 have been reported as sensitive markers for metaplastic carcinomas [10]. As other members of the TNBC family, >90 % of metaplastic carcinomas are negative for ER, PR, and HER2.

Metaplastic carcinoma are thought to arise from altered epithelial and/or myoepithelial cells. This theory is supported by cytogenetic and molecular studies that demonstrate the same clonality in both the glandular and non-glandular components of the tumor, indicating a common stem cell origin [11, 12]. Collectively, metaplastic carcinomas fall into the category of basal-like breast cancer, however, recent studies suggested that a subset of these tumors displays transcriptomic features consistent with cells undergoing epithelial-to-mesenchymal transition. These are referred to as Claudin-low tumors. Metaplastic carcinomas and claudin-low tumors are shown by comparative genomic hybridization to have low expression of GATA3-regulated genes and of genes responsible for cell-cell adhesion with enrichment for markers linked to stem cell function [13].

A typical apocrine carcinoma shows diffuse positivity for GCDFP-15 [14], and BCL-2 negativity. Staining for ER, PR is usually negative. A portion of these tumors is also negative for HER2 protein over-expression (triple negative). Androgen receptor expression in ER-negative breast tumors was found to be associated with apocrine differentiation [15].

The immunophenotypic signature described above has inspired researchers to look for a similar “apocrine molecular signature.” Microarray studies show increased androgen signaling and overlap with the HER2 group of tumors. However, this proposed molecular signature does not correlate well with apocrine morphology. Approximately, only half of carcinomas with apocrine differentiation show this molecular signature. Comparative genomic hybridization has identified several copy number alterations in carcinomas with apocrine differentiation including gains of 1p, 1q and 2q, as well as losses of 1p, 12q, 16q, 17q, and 22q [16]. However, these are also common alteration regions that are seen in breast carcinoma in general. This data suggests that although carcinomas with apocrine differentiation may have a characteristic morphology and immunophenotype, they do not represent a distinct molecular entity.

In the original series by Silver and Tavassoli, all tumors showed strong, diffuse positivity for pan-cytokeratin and CAM 5.2, which is useful in differentiating these tumors from sarcomas. EMA was positive in areas of conventional ductal carcinoma, but was very weak and focal in the multinucleated tumor cells. All tumors were negative for ER and PR. HER-2 was also negative in the majority of cases, especially those with node-negative disease [17]. P53 expression was present in 71 %, but none expressed bcl-2. Ki-67 proliferation index was also increased with a mean of 33 %.

The data on pleomorphic carcinoma is very sparse due to the rarity of these tumors. Nevertheless, the majority of these tumors show aneuploid DNA content and high S-phase.

The tumor cells are negative for ER, PR, and HER-2, and frequently positive for epithelial membrane antigen (EMA), S-100 protein, and E-cadherin (Fig. 11.2d). Expression of basal cytokeratins (CK 5/6 and 14) was also identified in five out of six cases in one study, suggesting that secretory carcinoma belongs to the basal-like group of breast cancer [18].

In 2002, Tognon et al. [19] have shown that secretory carcinoma is characteristically associated with t(12, 15) that results in ETV6-NTRK3 gene fusion, the same translocation which was originally described in congenital fibrosarcoma and cellular mesoblastic nephroma. Additionally, FISH analysis shows alteration of the ETV6 gene in both the in situ and invasive components [18].

The majority (>90 %) of MC are negative for ER, PR, and HER-2, with variable expression of basal cytokeratin (CK5/6), EGFR, and P53. The intense lymphocytic infiltrate is predominantly CD3+T lymphocytes. Not surprisingly, MC shows a high proliferation index with Ki-67.

MC heirs a lot of its molecular features from its basal-like family of breast cancers, including EGFR gene amplification, TP53 gene mutation, and increased incidence in patients with BRCA1 gene mutations [20]. Some questioned the role of Epstein-Barr virus infection in MC given its morphologic similarities with lymphoepithelial carcinomas of other organs. Whereas, one study showed an association between Epstein-Barr virus and MC, another study failed to reproduce this link [21, 22].

Expression of basal cytokeratins, particularly CK5/6 and CK14 is considered the sine-qua-non of BLBC (Fig. 11.3b). CK17 is also present in approximately 50 % of cases, but may be focal and weak. The expression of EGFR in BLBC varies in several studies, ranging from 45 to 70 %. Since more than 80 % of BRCA-1 associated cancers cluster in the basal-like category, it is not surprising that basal CKs and EGFR expression are also observed in BRCA-1 associated breast cancers. However, attempts to use these markers together with hormone receptors to predict mutation status in these patients has not been successful due to the high overlap between both BRCA-1 and non BRCA-1 associated BLBC [26]. Various other immunohistochemical markers have been studied as a tool to recognize and further characterize this specific subset of tumors. Insulin-like growth factor-II mRNA-binding protein 3 (IMP3), which was first introduced as a marker of aggressive behavior in renal cell and urothelial carcinomas [27], has been demonstrated in 78 % of TNBC, and correlates with CK5/6 expression (Fig. 11.3c) [28]. C-KIT has been reported in approximately 45 % of BLBC. P53 over-expression is also more common in BLBC compared to all breast cancers. Vascular endothelial growth factor (VEGF), maspin, osteopontin, integrin β4, caveolin1 and 2 have all been reported to be preferentially expressed in BLBC [29–33]. However, the only IHC signature of BLBC that has been validated by expression profiling demonstrates that a panel composed of ER, HER2, CK 5/6, and EGFR can identify these tumors with 100 % specificity and 76 % sensitivity [34].

The literature has been enriched by many studies that focus on better understanding of the molecular background of BLBC, in attempt to translate this molecular phenotype into targeted therapy. TP53 mutation has been identified in up to 83 % BLBC cases. The mutation is thought to be an early event in tumorigenesis and is related to poor prognosis and resistance to chemotherapy [35, 36]. The link with BRCA-1 gene mutation is well established. More than 80 % of BRCA-1 associated cancers cluster in the basal-like category [37], and many sporadic BLBC were shown to have altered BRCA1 activity and loss of function. Approximately 10–20 % of BLBC show methylation of gene promoter, and some have decreased BRCA1 mRNA. X-chromosome abnormalities, including defects in inactivation, were also identified. Additionally, the dominant-negative transcriptional regulator ID4 has been shown to regulate BRCA1 expression and to be preferentially expressed in BLBC [38–40]. Additionally, the loss of one TP53 allele in mice with mammary-specific deletion of BRCA1 dramatically accelerates mammary tumorigenesis, suggesting that TP53 mutations may act synergistically with BRCA1 defects in sporadic BLBC to drive tumor initiation.