Rare Breast Carcinomas: Adenoid Cystic Carcinoma, Neuroendocrine Carcinoma, Secretory Carcinoma, Carcinoma With Osteoclast-like Giant Cells, Lipid-Rich Carcinoma, and Glycogen-Rich Clear Cell Carcinoma

Clinical Presentation

Breast AdCC displays a clinical presentation that does not differ much from usual invasive breast carcinoma. The age at diagnosis ranges between 25 and 97 years (mean and median ages in the largest cohort were 63 and 62 years, respectively). The incidence ratio increases prominently at 35 to 44 years of age, with a less marked rise in incidence at older ages and an apparent plateau beginning at 55 to 64 years of age. White females appear to be at increased risk compared with Black women.

AdCC generally presents as a palpable, discrete, firm mass. A preferential subareolar/periareolar location has been suggested; however, in the largest series, the upper outer quadrant was predominantly affected (36%), whereas only 10% occurred in the nipple/central region. The lesions are equally distributed between the two breasts, and multifocal or bilateral tumors seem to be rare. The tumor is rarely fixed to the overlying skin, nipple, or pectoral muscles. Nipple discharge is not a common symptom; neither is pain or tenderness. The latter, when present, has not been particularly correlated with the presence of perineural invasion histologically, which is a rather rare finding.

Clinical Imaging

Mammary AdCCs usually do not show the classic stellate-shaped appearance of IDC-NSTs, but no specific imaging features have been described for this histological type of breast cancer. Mammographically, the lesions may appear as developing asymmetrical densities, irregular masses, or well-defined rounded nodules. Calcifications are rarely observed. On ultrasound examination, AdCCs display irregular, heterogeneous, or hypoechogenic features, with minimal vascularity on color Doppler imaging. Magnetic resonance imaging (MRI) may help better delineate the extent of tumor and thus avoid positive margins when breast conserving surgery (BCS) is indicated. A particular pattern of enhancement on MRI has been described and is reported to be associated with a conspicuous stromal component.

Gross Pathology

At gross examination, AdCCs are predominantly described as well-defined lesions, with rounded or lobulated borders. Cystic structures may be evident, as well as cystic degeneration in large lesions. Tumor size ranges from 0.1 to 16.0 cm (mean, 2.1 cm; median, 1.8 cm) and seems to be associated with histological grade.

Microscopic Pathology

AdCCs of the breast are histologically indistinguishable from examples in other sites. Although often circumscribed at imaging and gross examination, all types of breast AdCCs typically display a microscopic infiltrative growth pattern, in which the tumor infiltrates the surrounding breast tissue beyond a central grossly apparent nodule ( Fig. 30.1 ). They are composed of a dual cell population ( Fig. 30.2 ), characterized by production of mucinous and basement membrane material, and may exhibit tubular, trabecular, cribriform, and/or solid patterns of growth. In most cases, a mixture of different growth patterns is observed ( Fig. 30.3 ). The morphological heterogeneity of AdCCs has been well documented and should be taken into account when performing histological analysis of small samples such as core or vacuum-assisted biopsies (VABs) of breast tumors. It is unusual to witness an intraductal component.

The classic-type AdCC typically has a cribriform, trabecular, or tubular growth pattern and features variably sized and usually smoothly contoured islands of predominantly basaloid neoplastic cells arranged to compose pseudolumina and true glandular spaces, giving rise to a sieve-like appearance ( Fig. 30.4 ). The nuclear grade tends to be low to intermediate and mitotic figures are rare. More solid areas can be appreciated although care should be taken not to confuse with the solid-basaloid variant. The pseudolumina represent the vast majority of spaces and are actually stromal invaginations, often containing eosinophilic hyaline material (periodic acid-Schiff [PAS] positive, diastase resistant) and/or lightly basophilic myxoid substance (alcian blue positive). Ultrastructural analysis has demonstrated that these materials are duplicated basal lamina and glycosaminoglycans. True glandular lumina are far less prevalent and are lined by cuboidal cells with more abundant and eosinophilic cytoplasm. Classic AdCC is characterized by rounded or elongated tubules lined by epithelial cells and surrounded by single or multiple layers of basaloid cells ( Fig. 30.5 ). However, small nests/trabeculae or sheets of closely packed basaloid cells, with few or even no pseudocystic and true glandular spaces can be appreciated ( Figs. 30.6 and 30.7 ). Sebaceous differentiation and squamous metaplasia ( Fig. 30.8 ) can be found, as well as adenomyoepitheliomatous ( Fig. 30.9 ) and syringomatous areas, indicating some structural similarities with adenomyoepitheliomas and low-grade adenosquamous carcinomas of the breast.

In 2002, Shin and Rosen described the more aggressive SB-AdCC variant which is composed of large nests of basaloid cells, with pronounced nuclear atypia, frequent mitotic figures, and necrosis. In addition, perineural invasion is frequently identified. These tumors are composed predominantly of an epithelial cell population associated with decreased myoepithelial differentiation. However, they do maintain the MYB protein overexpression similar to the more common type of ACC. This subtype is important to identify as more recent studies have suggested this subtype has a greater tendency to develop axillary node metastasis as well as higher rates of local and distant recurrences. Although there are limited data on this subgroup, studies suggest that chemotherapy as well as radiotherapy should be considered.

A third very rare subtype of AdCC of the breast has also been described similar to that seen in the salivary gland in which the AdCC is associated with a high-grade carcinoma. This is termed AdCC with high-grade transformation and is defined by the presence of classic AdCC with multiple areas of differentiation including metaplastic carcinoma, small cell carcinoma, high-grade IDC, myoepithelial carcinoma, and malignant adenomyoepithelioma. DNA analysis showed similar molecular alterations between AdCC and the other components, suggesting that AdCC neoplastic cells can acquire aggressive potential. Although there are only scattered reports of these cases in the literature, it seems reasonable that more aggressive therapy is required for this subtype.

It is recommended to routinely grade AdCCs following the Nottingham grading system. Most classic-type tumors display mild to moderate nuclear pleomorphism and low to moderate mitotic activity and are classified as histological grade 1 or 2, depending on the proportion of solid areas. However, SB-AdCC can show marked nuclear atypia and is characterized by large, solid nests which typically yields to a Nottingham grade 3 designation.

Immunoprofile

Breast AdCCs are generally negative for estrogen receptor (ER), progesterone receptor (PgR), and HER2 expression. Whereas rare cases morphologically reviewed to confirm the diagnosis of AdCC showed focal positivity for ER and less frequently for PgR, HER2 overexpression and/or HER2 gene amplification have not been described. Taken together, it is reasonable to conclude that the vast majority of breast AdCCs display a triple-negative phenotype. Moreover, AdCCs frequently, if not always, express basal-like markers such as c-kit, p63, epidermal growth factor receptor (EGFR), and high molecular weight cytokeratins (CKs), being therefore classified as part of the basal-like molecular phenotype according to a validated immunohistochemical (IHC) surrogate panel. Proliferative ratios as defined by the Ki-67 index are variable, with levels ranging from 4% to 70% in one series, and may not be associated with prognosis.

The two distinct cell populations of AdCCs are best appreciated by IHC. The basaloid cells express p63, basal CKs such as CK14 and CK17, vimentin, S100 protein, actin, and calponin, whereas the epithelial cuboidal cells lining the true glandular lumina show strong positivity for c-kit, luminal CKs including CK7 and CK8/18, carcinoembryonic antigen (CEA), and epithelial membrane antigen (EMA). Furthermore, the stromal hyaline material can be highlighted by staining for collagen IV and laminin.

Molecular Pathology

Although microarrays have been extensively applied to the study of breast cancer, most analyses, including the seminal studies by Perou and Sorlie and coworkers, have not included representative numbers of special histological types of breast cancer. Therefore, the so-called molecular taxonomy of breast cancer (i.e., luminal A, luminal B, HER2, basal-like, and normal breast–like molecular subtypes) has been derived from analysis of only IDC-NST and few invasive lobular carcinomas. Weigelt and colleagues directly investigated the transcriptome of 11 histological special types of breast cancer, including AdCCs. Unsupervised hierarchical cluster analysis revealed that AdCCs clustered together with metaplastic carcinomas and that these tumors displayed a basal-like phenotype, corroborating the basal-like IHC profile of these tumors (i.e., lack of ER, PgR, and HER2 expression; low levels of CK19, androgen receptor [AR], and CK8/18; and high levels of c-kit, vimentin, S100, CK14, and CK5/6 expression). In addition, molecular subtype analysis with a single sample predictor showed that two AdCCs were of the basal-like phenotype, whereas two AdCCs were of the normal breast-like phenotype, a molecular subtype that is currently considered to be an artifact of sample representation (i.e., high content of normal tissue contamination). These data also illustrate the heterogeneity of the basal-like phenotype, which is predominantly composed of high-grade IDC-NST and metaplastic carcinomas but also includes a subgroup of breast cancers with low-grade histology and an indolent behavior, such as AdCCs and secretory carcinomas (see later). Importantly, however, there is evidence to suggest that in contrast to high-grade IDC-NST of the basal-like phenotype, AdCCs do not display BRCA1 downregulation, which may in part explain the differences in prognosis. These observations highlight the importance of histological subtyping of breast cancer, given that although a subgroup of IDC-NST and AdCCs displays the triple-negative/basal-like phenotype, the management of patients with IDC-NST is fundamentally different from those with AdCCs.

At the genomic level, AdCCs rarely display aneuploidy and often harbor simple genomes, indicating low levels of genetic instability. Frequent copy number alterations detected by microarray comparative genomic hybridization include focal gains of 1p, 11p, 12p, 16p, and 19p and focal losses on 6q and 9p. Of note, AdCCs do not harbor concurrent gains of 1q and 16p and losses of 16q as typically do low-grade IDC-NSTs. Moreover, breast AdCCs significantly differ at the genomic level from other basal-like IDC-NSTs.

In a way akin to secretory carcinomas, another form of indolent triple-negative and basal-like disease, AdCCs are also characterized by a recurrent specific translocation. Although the existence of t(6;9)(q22–23;p23–24) as characteristic of salivary gland tumors was known for many years, it was only in 2010 that Persson and associates reported that this translocation was found in all AdCCs analyzed (breast, salivary, ceruminal, and lacrimal glands) and that it leads to the fusion of the oncogene MYB (6q22-q23) with the transcription factor NFIB (9p23-p24). Although distinct breakpoints and fusion transcripts were reported, the common denominator of these rearrangements is the deletion of a microRNA target site of MYB, ultimately resulting in MYB overexpression. Subsequent analyses have confirmed that the translocation t(6;9)(q22–23;p23–24) is specific of AdCCs in the context of salivary gland tumors, but the prevalence seems to be lower than initially suggested. The MYB-NFIB fusion has been found in about one-third to one-half of salivary gland AdCCs tested; however, MYB overexpression as defined by IHC or quantitative reverse transcription polymerase chain reaction (qRT-PCR) was far more prevalent. Wetterskog and coworkers studied using fluorescence in situ hybridization (FISH) 13 breast AdCCs, of which one did not harbor the MYB-NFIB fusion gene. Nevertheless, qRT-PCR analysis revealed that all cases, including the fusion-negative case, displayed MYB overexpression as compared with histological grade–matched IDC-NST and basal-like IDC-NST. Those results provide strong circumstantial evidence to suggest that MYB overexpression is a molecular driver of AdCCs, is often, but not always, underpinned by t(6;9)(q22–23;p23–24), and may be a novel specific therapeutic target.

Additional potential targets have been reported on the basis of the analysis of an AdCC metastatic to the kidney. PTEN and PIK3CA mutations were found in both primary and metastatic lesions, possibly explaining the more aggressive behavior displayed by this case. Although PIK3CA and PTEN mutations may be more prevalent in breast cancers of the basal-like subtype, such as metaplastic breast cancers, more recent examination of the genomic landscape of AdCC shows that most tumors lack TP53 and PIK3CA mutations, and most show an array of somatic mutations including MYB, BRAF, FBXW7, SMARCA5, SF3B1, and fibroblast growth factor receptor 2 (FGFR2). The SB-AdCC variant also shows CREBBP mutations and NOTCH activating gene mutations which may provide potential therapeutic targets in the future. Overall, the mutational burden and mutational repertoire of breast AdCCs are more similar to those of salivary gland AdCCs than to those of other types of triple-negative breast cancers.

Treatment and Prognosis

The prognosis of AdCC depends upon the histological subtype and the subtype must be taken into consideration in determining therapeutic options. In contrast to salivary gland AdCCs, mammary classic-type AdCCs are characterized by an extremely good prognosis even though they are typically triple-negative, with slow progression and near-absence of lymph node metastasis. In rare cases, focal positivity for ER and PgR has been described, but the therapeutic significance of this is unknown. Local recurrence may occur if inadequately resected (≤56%). In the largest cohort, the rate of lymph node metastasis was 2.6%. Distant metastases, albeit rare, have been reported, mostly to the lung. Overall 5-year and 15-year survival rates are 98.1% and greater than 90%, respectively. Considering the indolent behavior of classic-type AdCC of the breast, surgical excision with clear margins alone may be an adequate treatment. BCS may be favored, although one must be aware that high rates (≤37%) of positive margins occur owing to the very infiltrative margins of the lesions. Given that classic-type AdCCs rarely metastasize to axillary lymph nodes and that distant metastasis may occur in the absence of lymph node metastasis, the role of axillary lymph node dissection (including sentinel lymph node biopsy) has been questioned, if not abandoned, in this context. Although the use of adjuvant radiation therapy has been described, its usefulness is yet to be fully defined.

SB-AdCC has been shown to have a greater tendency to develop axillary node metastasis as well as higher rates of local and distant recurrences to bone, lung, and skin. Cases from prior series have been treated with chemotherapy and radiotherapy and these modalities should be considered in the SB-AdCC variant.

Differential Diagnosis

The main differential diagnoses of breast AdCCs are other breast carcinomas with a cribriform pattern of growth, both invasive and intraductal. In fact, in the study by Sumpio and colleagues, half of the misclassified cases were invasive ductal carcinomas with a prominent cribriform intraductal component. Cribriform carcinomas are composed of only one cell type, usually with more abundant and eosinophilic cytoplasm, and display glandular lumina with no mucinous or basement membrane material. In addition, breast carcinomas with a cribriform pattern (either invasive or in situ cribriform carcinomas, low-grade IDC-NSTs, or papillary carcinomas) almost invariably are ER+ and PgR+. Other IHC markers indicating myoepithelial differentiation may also be useful, such as p63 and high molecular weight CKs, which are expressed in neoplastic cells of AdCCs and not in cribriform carcinomas. In addition, c-kit, which is expressed in almost all AdCCs, has been proposed as an ancillary tool for differentiating AdCCs from other breast carcinomas.

Another potential pitfall is collagenous spherulosis, which is usually found in association with papillary proliferations and hyperplasia of usual type. In this context, reliance on p63 or smooth muscle actin (SMA) alone may prove misleading because these markers are expressed in both lesions. Rabban and colleagues have directly addressed this issue and demonstrated that the use of calponin and smooth muscle myosin heavy chain (SMM-HC), markers more specific of the myoid apparatus of myoepithelial cells (only expressed in collagenous spherulosis) and c-kit (expressed only in AdCC), may potentially be helpful in this context.

AdCCs with a predominant nonclassic pattern of growth, in particular the ones with solid and basaloid elements, may be difficult to differentiate from other high-grade breast carcinomas. IHC with low and high molecular weight CKs may be useful to highlight the dual cell population in AdCCs. In addition, CK7 allows one to identify the few epithelial cells lining true glandular lumina. p63, EGFR, and c-kit may not be of use because they may also be expressed in high-grade IDC-NST. Moreover, p63 is usually not expressed in cells with basaloid features. Finally, in this context, the use of FISH with split-apart or fusion probes to detect the t(6;9) rearrangement and qRT-PCR for the MYB-NFIB may play a role to establish the diagnosis of AdCC. In addition, MYB IHC has recently been shown to potentially be more sensitive and specific for breast AdCC than MYB labeling by FISH.

Clinical Presentation

No specific clinical presentation has been described for neuroendocrine breast carcinomas. Endocrine hormone–related syndromes are exceptionally rare in the breast. The age at diagnosis appears to be higher than that of patients with IDC-NST or lobular carcinomas. Most patients are postmenopausal women, in their sixth or seventh decade of life, although neuroendocrine differentiation can also occur in male carcinomas. Most patients present with a palpable nodule. Rapid growth and advanced stage at presentation are characteristic of small cell carcinomas.

Clinical Imaging

Neuroendocrine tumors generally appear as a circumscribed mass on mammographic and ultrasound examination.

Gross Pathology

Tumors with grossly well-defined or invasive borders can occur. When present, mucin production can be detected at gross examination, giving a soft and gelatinous appearance to the lesions. Tumor size ranged from 0.8 to 13.5 cm (mean, 2.7 cm; median, 2.2 cm) in one series.

Microscopic Pathology

The WHO classification recognizes two morphological variants of neuroendocrine neoplasms: neuroendocrine tumors and neuroendocrine carcinomas. NETs correspond in a way to well-differentiated neuroendocrine tumors of the lung and gut. Typically, the solid tumors are composed of densely cellular, solid nests and trabeculae of cells separated by delicate fibrovascular stroma ( Fig. 30.10 ). An alveolar pattern is frequently found, bearing resemblance to the alveolar variant of invasive lobular carcinomas. In a minority of cases, a carcinoid-like pattern with rosette-like structures and peripheral palisading can be observed ( Fig. 30.11 ). Tumor cells most often display a large cuboidal, spindle, or plasmacytoid shape with granular and eosinophilic cytoplasm ( Figs. 30.12 and 30.13 ). The nuclei are typically low to intermediate grade and may or may not show the stippled “salt and pepper”–like chromatin. Mucin can be a prominent feature, although care should be taken not to diagnose the hypercellular variant of mucinous carcinoma as a neuroendocrine tumor. Overall these are Nottingham grade 1 or 2 invasive carcinomas, and when ductal carcinoma in situ (DCIS) is present, it typically shows similar morphological features.

Neuroendocrine carcinomas, on the other hand, are high-grade invasive carcinomas which show neuroendocrine differentiation. NECs include small cell carcinoma and large cell neuroendocrine carcinomas, both of which are morphologically similar to their counterparts in the lung. Small cell carcinoma of the breast shows densely packed hyperchromatic cells with scant cytoplasm and often with prominent crush artifact. Mitotic figures are abundant and areas of necrosis and lymphovascular invasion are frequently identified. This variant can be mixed with areas of poorly differentiated carcinoma, NST. However, it is important that this variant be differentiated from the others given that it is associated with an unfavorable prognosis and may be responsive to chemotherapy used for small cell carcinomas at other sites, especially when metastatic. Given the morphological and IHC overlap between small cell carcinomas of the breast and other anatomic site cancers (see “Differential Diagnosis”), the diagnosis of a primary small cell mammary carcinoma depends on the clinical exclusion of a nonmammary primary site and/or the histological finding of an in situ component. The latter most frequently displays high nuclear grade and cribriform or solid architecture and may be composed of cells that display cytological features and IHC profiles similar to those found in invasive carcinoma. A primary small cell mammary carcinoma arising from in situ and invasive lobular carcinoma is on record, although evidence of clonality between the two lesions to distinguish it from a collision tumor was not provided.

Large cell neuroendocrine carcinoma of the breast is morphologically similar to that seen in the lung. These are poorly differentiated tumors composed of clusters of large cells, with moderate to abundant cytoplasm, pleomorphic nuclei with vesicular to finely granular chromatin, and high mitotic activity. Similar to its lung counterpart, neuroendocrine differentiation in large cell carcinomas of the breast is demonstrated by IHC and can often be diagnosed as poorly differentiated carcinomas (NST) if the neuroendocrine features are not appreciated.

Histological grading of neuroendocrine carcinomas should follow the Nottingham grading system. Given the low to intermediate nuclear pleomorphism and the predominant solid architecture, most neuroendocrine tumors of the breast are classified as histological grade 1 or 2 (45% and 40%, respectively) depending on the mitotic activity, which has varied from 4 to 45 per 10 high-power fields (HPFs). Small cell and large cell variants are uniformly classified as grade 3 and should be diagnosed as neuroendocrine carcinomas. Although Sapino and associates demonstrated that histological grading was one of the most important prognostic parameters in neuroendocrine carcinomas, Tian and coworkers reported that neither mitosis counting nor histological grade predicted survival in a cohort of breast neuroendocrine carcinomas. Given the retrospective nature of these studies and the fact that patients received different modalities of systemic therapy, further studies are warranted to determine the real impact of histological grading on the outcome of patients with breast neuroendocrine carcinomas. Despite the controversies, histological grading with the Nottingham grading system is recommended.

Immunoprofile

Chromogranin A and synaptophysin have been considered the most sensitive and specific neuroendocrine markers in breast pathology; however, CD56 can also be used to demonstrate neuroendocrine differentiation, although it is less specific ( Fig. 30.14 ). Insulinoma associated protein 1 (INSM1) is a transcription factor that has been shown to be a sensitive and specific marker for neuroendocrine tumors from various sites. INSM1 leads to nuclear staining as opposed to the other neuroendocrine markers which are cytoplasmic. Although included in the assessment of neuroendocrine differentiation in some studies, expression of neuron-specific enolase should not be used to support a diagnosis of neuroendocrine carcinoma. ^79. Although older classification schemes required that more than 50% of the tumor express neuroendocrine markers, there is currently no set threshold for neuroendocrine expression in a tumor to be labeled a neuroendocrine neoplasm. The 5th Edition of the WHO series states the tumor must have neuroendocrine morphology and a diffuse and uniform expression of neuroendocrine markers. It should be noted these carcinomas often have areas with more typical IDC-NST in addition to areas with neuroendocrine differentiation, so the entire tumor may not have the same appearance or IHC profile.

The immunophenotype of mammary neuroendocrine carcinomas varies according to the morphological variant. The majority of low- to intermediate-grade neuroendocrine tumors are HR+ and HER2−. In one series comprising 74 tumors, 95% were ER+, 80% were PgR+, and 91% were HER2−. GATA3 is reported to be positive in greater than 95% of cases. Although neuroendocrine tumors of the breast are not graded based on the Ki-67 proliferation index and/or the presence of necrosis as are NETs in other organ systems, the number of mitoses remains the main parameter-influencing grade. Some studies have suggested that proliferative activity as defined by Ki-67 IHC may be a significant independent prognostic factor.

In contrast, a lower proportion of neuroendocrine carcinomas express HR. The largest series describes 67% and 56% of positivity for ER and PgR, respectively. HER2 overexpression and HER2 gene amplification appear to be vanishingly rare in these cancers. A high Ki-67 index is typically found as are areas of necrosis. Moreover, breast neuroendocrine carcinomas are typically positive for GATA3, CKs (including CK7 and cell adhesion molecule 5.2 [CAM5.2]), bcl-2 (small cell variant), and E-Cadherin (ECAD). On a cautionary note, thyroid transcription factor-1 (TTF-1) is expressed in a proportion of mammary small cell carcinomas, limiting its use for defining primary tumor site in this context.

Molecular Pathology

Neuroendocrine neoplasms of the breast are most often of the luminal subtype and are genetically heterogeneous. However, neuroendocrine neoplasms are distinct from the genetic alterations typically found in IC-NST. Compared to ER+, HER2− IC-NST, neuroendocrine neoplasms are typically enriched for mutations in TBX3 and FOXA1 while they harbor a lower frequency of TP53 and PIK3CA mutations. ESR1 and BCL2 have also been found to be upregulated in neuroendocrine neoplasms compared with IC-NST. Mutations in GATA3 and ARID1 are also frequently identified. Neuroendocrine neoplasms also differ in their copy number profile compared with IC-NST as they show a low frequency of concurrent 1q gains and 16q losses as is typically seen in ER+ breast carcinoma NST. However, both common forms of IC-NST and neuroendocrine neoplasms harbor amplifications such as 8p11.23-11.21, 8q24.3, and 8q24.12.

Ang and associates described potentially actionable discrete targets in one-third of 15 neuroendocrine neoplasms in their study. There were three patients with PIK3CA exon 9 E542K mutations, two of which also harbored point mutations in FGFR family members (FGFR1 P126S, FGFR4 V550M). Single mutations were also found in each of KDR (A1065T) and HRAS (G12A). However, FGFR and RAS family mutations are exceedingly rare in other breast cancers. Likewise, activating mutations in the receptor tyrosine kinase KDR (VEGFR2), reported in angiosarcomas and non–small cell lung cancers, may be sensitive to VEGFR kinase inhibitors.

Treatment and Prognosis

There are multiple conflicting reports regarding the outcome of low-grade neuroendocrine tumors. Initial publications have suggested that non–small cell neuroendocrine tumors would have a better prognosis than unselected breast carcinomas. In a more recent study, Wei and associates analyzed 68 cases treated at the MD Anderson Cancer Center and described that, despite similar age and disease stages at presentation, neuroendocrine carcinomas showed a more aggressive course than IDC-NSTs, with a higher propensity for local and distant recurrence and poorer overall survival. The 5-year overall survival rate and disease-free survival rate were 84% and 65%, respectively. High nuclear grade, large tumor size, and regional lymph node metastasis were significantly associated with survival. When corrected for age, tumor size, histological grade, and ER status, neuroendocrine tumors have been found to be associated with a worse clinical course. These observed differences may be a result of distinct case selection and treatment modalities. Overall, staging and histological grading remain the most important prognostic factors, and there are no specific recommendations differing from IC-NSTs for the treatment of patients with neuroendocrine tumors. Those patients must be treated according to general guidelines for breast cancer.

As for neuroendocrine carcinoma (small and large cell variants), there is no debate that it should be considered a high-grade cancer with unfavorable prognosis. Lymph node metastasis is reported in one-third of cases. A large Surveillance, Epidemiology, and End Results (SEER) data-based study on the small cell variant revealed these carcinomas have a shorter disease-specific survival (DSS) and overall survival (OS) compared to other breast neuroendocrine tumors. Nevertheless, one study indicates that low-stage breast small cell carcinomas respond to conventional treatment without progression of the disease at a follow-up of 33 to 48 months. Moreover, compared with lung small cell carcinoma, breast neuroendocrine carcinomas with local or regional disease do have a more favorable outcome. However, similar to lung small cell carcinoma, breast neuroendocrine carcinomas with distant disease have a very poor prognosis. Currently, the management of neuroendocrine carcinomas also does not differ from that of IC-NST of the breast, and hormonal therapy is indicated in patients with ER+ tumors. However, in the metastatic setting, therapies used for small cell carcinoma of the lung, such as combinations of platinum compounds and etoposide, have shown some benefit.

Differential Diagnosis

Neuroendocrine carcinomas of the breast must be initially distinguished from usual invasive carcinomas with focal neuroendocrine differentiation. Adoption of the WHO criteria (i.e., morphology of neuroendocrine differentiation and diffuse and uniform immunoreactivity for neuroendocrine markers) is required. Furthermore, primary breast neuroendocrine carcinomas of all variants must be differentiated from metastasis of neuroendocrine tumors of other anatomic sites. The presence of an in situ component, ideally displaying similar cytological features, may be considered the best evidence in favor of a breast primary. IHC should be evaluated with caution because HR expression may occur in nonmammary neuroendocrine tumors, in particular PgR expression. Likewise, antibodies for site-specific transcription factors, such as TTF-1 and CDX2, are of limited use in tumors with neuroendocrine differentiation. For instance, TTF-1 has been shown to be expressed in 20% of mammary small cell carcinomas, whereas the entire spectrum of lung neuroendocrine neoplasms may display focal to diffuse ER and PgR expression. In this context, the expression of gross cystic disease fluid protein-15 (GCDFP-15) and/or mammaglobin may indicate a breast primary because nonmammary neuroendocrine tumors failed to express both markers in one study. GATA3 expression may also be useful as this marker is typically seen in those neoplasms originating in the breast. Finally, before rendering a diagnosis of a breast neuroendocrine carcinoma without an in situ component, one may best rule out any evidence of a neuroendocrine neoplasm in another anatomic site.