Introduction

Neuroendocrine carcinoma of the breast (NECB) is a rare form of breast cancer accounting for less than 0.5 % of all cases [1, 2]. While the majority of neuroendocrine tumors arise from neuroendocrine cells in the bronchopulmonary system, gastrointestinal tract, and pancreas, NECB is thought to arise from the divergent differentiation of neoplastic epithelial cells during breast carcinogenesis, rather than preexisting neuroendocrine stem cells. This theory is supported by studies demonstrating that most NECBs have a similar histologic appearance to normal-type breast carcinoma, undifferentiated breast cancer cells are capable of expressing neuroendocrine markers, and the fact that benign neuroendocrine tumors have never been reported in the breast [3–7]. The World Health Organization (WHO) classifies NECB into three subtypes: well-differentiated neuroendocrine tumors, poorly differentiated neuroendocrine carcinoma or small cell carcinoma, and invasive breast carcinoma with neuroendocrine differentiation [8]. Implicit in the definition of a neuroendocrine carcinoma of the breast, is the presence of neuroendocrine markers in more than 50 % of cells [9].

Because of its rarity, the current understanding of NECB is limited to case reports and small case series. Previous reports emphasized the aggressive nature and poor prognosis of NECB. However, these studies focused primarily on poorly differentiated NECB [4, 10, 11]. More recently, a systematic review of the literature identified only 108 reported cases of NECB but lacked information on neuroendocrine classification [12]. A population-based analysis of NECB with long-term outcomes is lacking. The purpose of this study, therefore, was to utilize the surveillance, epidemiology, and end results (SEER) database to study a large cohort of NECB, specifically focusing on the histopathological subtypes and prognosis.

Methods

The SEER database was used to identify all patients with a primary NECB. Since the WHO first defined NECB in 2003, only cases from 2003 to 2010 were included. The 2012 WHO classification system was utilized for categorizing and abstracting NECB cases based on the following histopathology: well-differentiated neuroendocrine tumor (SEER code 8246, Fig. 1), poorly differentiated or small/oat cell neuroendocrine tumor (8041–8045), and adenocarcinoma with neuroendocrine features (ANF, 8574). Although not currently listed in the WHO classification system for NECB, also included were carcinoid (8240) and large cell neuroendocrine (8013) tumors.

Fig. 1
figure 1

An example of a well-differentiated neuroendocrine tumor of the breast. a H&E stain (×100) demonstrates the solid to trabecular growth pattern often seen with neuroendocrine differentiation. b The nuclear features of neuroendocrine differentiation are apparent at higher power (×200) with granular, speckled chromatin. c Diffuse staining with synaptophysin. and d chromogranin antibodies is present and characteristic (×100)

Full size image

Descriptive statistics were calculated for demographic (age, gender, race), clinicopathologic (histological subtype, grade, TNM stage, estrogen (ER), and progesterone (PR) receptor status), and treatment (surgery of primary, lymph node surgery, radiation) characteristics. Survival analysis was performed for both disease specific (DSS) and overall (OS) survival using the Kaplan–Meier method. Statistical significance was assessed using the Mantel–Cox log-rank test. Data were unadjusted for demographic, tumor-related, or treatment variables. Figures were created using Graphpad Prism 6.0 (Graphpad Software, Inc; La Jolla, CA).

Univariate and multivariate Cox proportional hazards models were then created to evaluate factors associated with DSS and OS. Independent variables included in the models were age, race, gender, histological subtype, grade, ER/PR status, T, N and M stages, surgery type, receipt of radiation, and whether lymph node sampling was performed. Results are reported as Odds ratio (OR) with 95 % confidence intervals (CI). Statistics were performed via STATA-MP 11.2 (StataCorp LP; College Station, TX).

Results

Between 2003 and 2010, 284 cases of NECB were recorded in the SEER database: 148 (52.1 %) as well-differentiated neuroendocrine carcinoma, 73 (25.7 %) small cell carcinomas, 42 (14.8 %) with ANF, 14 (4.9 %) large cell neuroendocrine carcinomas, and 7 (2.5 %) carcinoid tumors. Cohort characteristics are listed in Table 1. As expected, most patients were female (96.8 %) and white (81.7 %) while there was a balanced age distribution. A high percentage of tumors (37.3 %) were graded as poorly differentiated though many still expressed ER (46.5 %) and PR (35.6 %) receptors. Many patients had advanced stage disease: 36.2 % presenting with regional lymph node metastases and 20.4 % with systemic metastases. Overall cancer stage was similar across all histologic subtypes, except stage 1 disease predominated in well-differentiated and ANF subtypes (Table 2). Primary surgery was mastectomy in 35.2 % and lumpectomy in 36.6 %, while 27.8 % did not undergo surgery. The latter is congruent with the proportion of patients presenting with stage IV or stage unknown disease.

Table 1 Demographic, clinicopathologic, and treatment characteristics for all patients with neuroendocrine carcinoma of the breast
Full size table
Table 2 Stage stratification for primary neuroendocrine carcinomas based on histological subtype
Full size table

Stage-stratified DSS and OS survival curves are presented in Fig. 2. Five-year DSS rates for stage I–IV NECB were 88.1, 67.8, 60.5, and 12.4 %, respectively (p < 0.0001) while 5-year OS rates were 77.9, 57.3, 52.9, and 8.9 %, respectively (p < 0.0001). When stratified by histology, long-term outcomes were significantly more favorable for well-differentiated neuroendocrine carcinomas and ANFs compared to small cell carcinomas, large cell carcinomas, and carcinoids (Fig. 3). Five-year DSS rates for well-differentiated neuroendocrine tumors, ANFs, small cell carcinomas, and large cell/carcinoids (combined because of the small numbers) were 74.0, 73.3, 50.5, and 49.1 %, respectively, while 5-year OS rates were 62.4, 68.9, 32.2. and 24.8 %, respectively.

Fig. 2
figure 2

Stage-stratified disease specific (a) and overall (b) survival for patients with primary neuroendocrine carcinoma of the breast

Full size image
Fig. 3
figure 3

Disease specific (a) and overall (b) survival for patients with neuroendocrine carcinoma of the breast stratified by histologic subtype: well-differentiated neuroendocrine carcinoma, small cell neuroendocrine carcinoma, adenocarcinoma with neuroendocrine features (ANF), and other primary neuroendocrine tumors

Full size image

On univariate Cox proportional hazards regression, small cell carcinoma was significantly associated with worse DSS (OR 1.97, 95 % CI 1.05–3.67) and OS (OR 2.66, 95 % CI 1.49–4.72) compared to other neuroendocrine tumors. Black race, poor differentiation, ER/PR receptor negativity, increasing T, N and M stage, and no primary breast surgery were all associated with worse DSS (Table 3). Age ≥ 80, poorly differentiated and undifferentiated tumors, ER/PR receptor negativity, advanced T, N and M stages, and no primary breast surgery were all associated with worse OS (Table 3). After controlling for other factors with multivariate Cox hazards regression analysis, the association of small cell carcinoma with DSS (OR 6.46, 95 % CI 0.88–47.68, p = 0.07) and OS (1.97, 95 % CI 0.47–8.22, p = 0.36) became less statistically significant.

Table 3 Univariate cox proportional hazards analysis for disease specific and overall survival
Full size table

Discussion

Feyrter and Hartmann first made the observation of breast tissue that resembled intestinal carcinoids in 1963 [13]. It was not until 1977, that Cubilia and Woodruff reported the first series of primary carcinoid tumors of the breast [14]. Since then, sporadic cases of NECB have been reported throughout the literature. Appreciation of NECB as a separate entity was greatly enhanced when the WHO included NECB in its 2003 classification of tumors report [15]. Despite the increased interest, few large series of NECB have been reported and information on long-term outcomes has been lacking [16].

The diagnosis of NECB can be challenging and depends on careful evaluation of core needle or excisional biopsy specimens. However, routine histological staining is made difficult in that many of the classic histopathological features of neuroendocrine carcinomas occurring in other organs are not present in NECB. Furthermore, mixed growth patterns with invasive ductal or lobular carcinoma, not otherwise specified (NOS) are often present. In one retrospective analysis, the diagnosis of NECB was not recognized initially in over two-thirds of the cases [17]. Histopathologic features concerning for neuroendocrine differentiation (most commonly papillary, nesting, or mixed growth patterns) should be confirmed by immunohistochemical staining with chromogranin, synaptophysin, neuron specific enolase, or other neuroendocrine markers [12, 17]. In addition, NECB should demonstrate an immunoprofile consistent with CK7 positivity and CK20 negativity similar to other breast cancers; nevertheless, there can be variability in this pattern. Interestingly, while many of these tumors demonstrate estrogen and progesterone receptor positivity, their expression is not diagnostic of NECB as ER/PR have been found expressed in other sites and are not universally expressed in breast carcinoma [18].

Because of their rarity as primary breast neoplasms, metastasis from another primary neuroendocrine tumor should be excluded, especially for small cell histology. Markers of pulmonary (TTF-1), GI (CDX-2), or pancreatic origin (PDX-1) should be negative. Careful clinical and radiographic evaluations should exclude other primary sites. Conversely, demonstration of an in situ carcinoma component is helpful for classification as a primary NECB. SEER makes every attempt to include only non-metastatic tumors in the primary site category. However, since this is secondary data, we are unable to conclude with complete certainty that no metastatic neuroendocrine neoplasms originating in other organs or tissues were included as breast primaries.

Within the context of NECB, three distinct subtypes have been described according to the WHO: well-differentiated neuroendocrine tumors, poorly differentiated neuroendocrine carcinoma, or small cell carcinoma, and invasive breast carcinoma with neuroendocrine differentiation [8]. Other rare histological subtypes have also been described: carcinoid [19], large cell neuroendocrine [20], endocrine mucoid carcinoma [21], and endocrine ductal carcinoma in situ [22]. Importantly, the impact of histological subtype on long-term outcomes has not been previously investigated. Our study utilized SEER, which advantageously codes these histological subtypes separately, to determine the prognostic value of histology in NECB. We found that small cell carcinomas, large cell carcinomas, and carcinoids behaved similarly and had significantly worse DSS and OS compared to well-differentiated neuroendocrine carcinomas and adenocarcinomas with neuroendocrine features.

Our study confirms previous findings that patients with NECB, in general, present with relatively more advanced disease than primary breast cancers [1]. In our study, 36 % of patients had lymph node metastases and 20 % presented with systemic metastases, significantly greater than patients with invasive ductal carcinoma (IDC, approximately 40 and 10 %, respectively) [23]. Patients with ANF were more likely to present with stage 1 disease, while the other histological subtypes presented at similar stages. Importantly, we found that NECB is associated with worse long-term outcomes compared to IDC [24, 25]. 5-year OS rates of 77.9, 57.3, 52.9, and 8.9 % for stage I–IV disease, respectively, are significantly higher than stage-stratified rates for IDC (100, 93, 72, and 22 %, respectively) [26]. Recently, a stage-matched comparative analysis demonstrated worse survival in NECB compared to IDC-NOS [27].

Our study highlights other demographic and histopathological characteristics with prognostic importance. Age, race, tumor grade, T, N and M staging and receipt of surgery, and lymph node sampling were all associated with survival. Despite its more aggressive behavior, treatment paradigms today are similar to that of IDC. Breast-conserving surgery and mastectomy were used with similar frequency in this population-based study [28–30]. Although the systemic treatment of early NECB typically follows that of IDC [12, 28], immunohistochemistry may be important for guiding adjuvant treatment regimens [31] as adjuvant endocrine therapies may offer potential in treating susceptible tumors [25]. For example, somatostatin analogs have been used for tumors with confirmed somatostatin-receptor expression [32].

The SEER database is a prospectively collected, population-based cancer registry, which captures 26 % of all cancer cases throughout the United States and provides impressive follow-up. Its large sample size enables the study of less-common diseases such as NECB, on which previous studies had been significantly limited by their small sample size. As with any registry-based analysis, inaccuracies in coding and abstracting may have occurred. In addition, no stringent prospective protocols for pathological review could be applied. Pathologist misinterpretation or misclassification of metastases in the breast from neuroendocrine tumors originating elsewhere is possible. Of note, SEER does not contain information on preoperative comorbidities, postoperative complications, margin status, use of adjuvant systemic therapies, all of which represent important variables in outcome analyses.

These data represent the largest series of NECB cases in the literature. We describe the epidemiological characteristics associated with this rare variant of breast cancer underscoring the advanced stage at presentation and the relatively poor prognosis of NECB. Small cell carcinoma subtype, in particular, is associated with worse DSS and OS compared to well-differentiated NECB and invasive carcinoma with neuroendocrine features. Future studies should consider reporting these subtypes separately based on the differences in prognosis.