Treatment Regimens and Survival Outcomes of Primary Neuroendocrine Carcinoma of Breast Compared with Inltrating Ductal Carcinoma: A Large Propensity Score Matching Study

Purpose The role of treatment regimens in Neuroendocrine carcinoma of breast (NECB) is not well dened. The aim of this study was to explore the clinical characteristics, treatment regimens and prognosis of NECB compared to inltrating ductal breast carcinomas(IDCB). Method NECB and IDCB patients diagnosed between January 1,2004 and December 12,2016 were enrolled in the researches from the SEER registries. A propensity score matching (PSM) with ratio of 1:5 was performed to correct the selection bias. The Kaplan-Meier curves was used in the analysis of variables. The Cox model was adopted to calculate univariate and multivariate analysis.


Introduction
Neuroendocrine carcinoma of breast (NECB) is a quite rare subtype of neuroendocrine neoplasia accounting for less than 1 % of all cases [1]. Neuroendocrine neoplasm is a heterogeneous kind of tumors which may originate from nearly all organs with endocrine system behavior [2]. About 70% of the neuroendocrine neoplasia occurred in gastroenteric pancreatic system, 25% in the bronchopulmonary system. Relative rare occurs in thyroid gland, skin, bladder and larynx [3].
Initially, NECB was rst reported by Feyrter at all in 1963 as a new kind of invasive breast cancer [4]. Then in 1977, Cubilla at all identi ed eight similar cases and summarized it characteristics [5]. Until 2002, the NECB was rstly de ned and in the following year 2003, it was adopted as an individual type of carcinoma by the World Health Organization (WHO) [6][7][8]. In previous, the speci c markers of NECB were de cient resulting in many NECB patients missed diagnosed in routinely clinical work.
In the WHO Classi cation of Tumors of the Breast of 2012, the classi cation of NECB was corrected and categorized as three subgroups:1) well-differentiated neuroendocrine tumors;2) poorly differentiated neuroendocrine carcinoma or small cell carcinoma of breast and 3) invasive breast carcinoma with neuroendocrine differentiation. Among the other type of breast cancers such as invasive ductal, lobular, colloid and papillary carcinoma, the neuroendocrine differentiation can also be observed [9].
No speci c therapeutic guidelines for the treatment of NCEB was published and the accessible data mainly derived from case reports and small scale retrospective trials. Most adopted strategies of NECB were referred to conventional breast cancer type. As regarding to the survival of NECB, the view was controversial: some researches indicated that the survival of NECB was poor, while others showed that it was better than in ltrating ductal breast carcinomas(IDCB) [6]. In addition, no signi cant difference between them was observed in some researches [10].
In this research, we present a propensity score matching (PSM) study with each NECB patients matching ve IDCB patients. The aim of this study was to explore the clinical characteristics, treatment regimens and prognosis of NECB compared to in ltrating ductal breast carcinomas (IDCB), for IDCB is the most popular subtye of malignant breast cancer.

Patients
The SEER Program maintained by the US National Cancer Institute (NCI) provided a data source for rare disease which contained 28% population of US, the largest publicly available cancer database [11]. NECB and IDCB patients diagnosed between January 1, 2014 and December 12, 2016 were concluded in the researches. Only primary neoplasm patients with de nitely pathologic diagnosis were eligible in the

Clinical Characteristics
Age at diagnosis, sex, race, primary site, marital status, seer stage (regional, localized, distant),TNM stage, molecular subtype (Luminal A, Luminal B, Triple Negative), treatment regimens (surgery, chemotherapy, radiation) of patients were collected from the SEER database and used to analyze the risk factors of NECB patients. Age were continuous variables and were classi ed as two categorical variables (< 60 year,≧ 60 year). Before 2015, the tumor stage was coded regarding to American Joint Committee on Cancer(AJCC) tumor, node, metastasis classi cation staging system,6th edition. From 2016, the category was according to 7th edition.
A PSM was performed between NECB and IDCB patients with 1:5 ratio by each variables including age at diagnosis, sex, race, primary site, marital status, seer stage (regional, localized, distant), TNM stage, molecular subtype (Luminal A, Luminal B, Triple Negative),treatment regimens (surgery, chemotherapy, radiation).

Statistical Analyses
Rates of overall survival (OS) and disease-speci c survival (DSS) were considered as endpoints of the survival of patients. OS was measured from the day of diagnosis to the day of death caused by any reasons. DSS was evaluated from the day of diagnosis to the day of death caused by breast carcinoma.
The Kaplan-Meier curves method was used in performing the survival analysis of clinical variables. The Cox proportional hazards model was adopted to calculate univariate and multivariate analysis. The outcome of risks of OS and DSS were reported as hazard ratios (HRs) with 95% con dence intervals (CIs). All the statistical analyses and graphics were performed with the SPSS 25.0 software (SPSS Inc, Chicago, Illinois, USA) and R statistical software (version-4.0.2.). P value < 0.05 is considered as statistically signi cant.

Patients
By researching the SEER database, a total of 698363 malignant mammary carcinoma patients was identi ed from January 1, 2004 to December 12, 2016. Of which ,426 NECB patients were collected and 295 patients nally enrolled in study. The demographic and clinical characteristics of eligible NECB and IDCB patients were shown in the Table 1 before the PSM. Signi cant differences were identi ed between NECB and IDCB patients population with regarding to age at diagnosis, sex, primary site, SEER stage, T stage, N stage, distant metastasis, molecular subtype, rate of surgical method, rate of chemotherapy and rate of radiation. But, there was no signi cant differences shown in race and marital status. Compared to IDCB patients, NECB patients were more elder (41.7% vs 53.2%, p < 0.0001). NECB patients presented with greater frequency of distant stage than IDCB patients(25.1% vs 5.9%, p < 0.0001). There was more triple negative type in NECB patients than that in IDCB patients(14.9% vs 6.4%, p < 0.001). The proportion of received surgery or radiation was less in NECB compared to those in IDCB, respectively. In contrast, Patients treated with chemotherapy in NECB was more than that in IDCB (52.9% vs 44.2%, p < 0.003).  1C and D). In the 3-year survival analysis, the rates of DSS of patients undergone surgery were over two times of patients who were not undergone surgery(73.6% vs 34.0%) ( Fig. 1E and F). To OS, the rates of patients undergone surgery were nearly three times of patients who were not undergone surgery (73.6% vs 25.7%). The 5-year rates were also with greater difference.

Survival Analysis After Psm
Each NECB patients were matched with ve IDCB patients data to gain more sight of baseline demographic and diagnostic differences by all variables as age at diagnosis, sex, race, marital status, primary site, SEER stage, T stage, N stage, distant metastasis, molecular subtype, rate of surgical method, chemotherapy and radiation. After matching, good balance with no signi cant difference was observed between the NECB and IDCB patients groups (Table 2).  The NECB patients demonstrated signi cantly poorer survival compared to IDCB patients in both DSS (p < 0.0001) and OS (P < 0.0001) (Fig. 2). Actuarial

Discussion
NECB is a very rare and aggressive subtype of malignant breast neoplasm with few respective clinical trials published [12][13][14]. The evidence for risk factors and treatment regimens were insu cient. Our studies is the rst and largest research to compare the NECB and IDCB by performing PSM to identify treatment regimens and clinical characteristics of NCEB. The current study demonstrated that NECB usually presents in white female population elder than 60, with advanced TNM stage. As previous studies reported, elder patients (> 60) were at a higher risk of NECB comparing to IDCB. In this research, it was shown that the advanced T stage and metastasis status were correlated with worse prognosis in NECB patients as expected. Chemotherapy and surgery could signi cantly improve the survival of NECB patients. Compared to IDCB patients, NECB patients were with poorer DSS and OS. Reference to molecular subtype, the proportion of luminal A type was the largest of all, followed by triple negative and luminal B type similar to the previous studies. The evidence for risk factors and treatment regimens were insu cient.
The prognosis of NECB versus usual type IDCB had been reported by several studies [15,16]. Some researchers showed that there was a signi cant difference between the two subtype. However, some considered that they had similar prognosis [16]. In this research, before matching, patients with NECB tend to be elder, at advance T stage, various proportion of molecular subtype and received different treatment regimens compared to IDCB groups. After matching, survival of NECB was signi cant worse than IDCB.
Estrogen receptor(ER) and/or progesterone receptor(PR) is positive in majority of NECB patients with the proportion of 68%, while, Her2 is usually negative [17]. Then the patients were subdivided as Luminal A and Luminal B through Ki-67 immunohistochemical staining(15%) [18,19]. The distribution of luminal A and B was equal identi ed by some studies. However, some studies discovered that the proportions of luminal A was larger than luminal B. And in some studies, luminal B type counting more [20,21]. Very rarely triple negative were observed [22]. In our research, the outcomes supported that luminal A type take the largest account of all, followed by triple negative and then luminal B type. Compared to IDCB,more proportion of NECB fall into triple negative group.
Regarding to the high grade malignancy and poor prognosis, e cient treatment is essential to NECB [23][24][25]. However, there was still no standard treatment for various stage NECB. Surgery is the rst choice in treating breast neoplasm [25]. As for general malignant mammary carcinoma,tumor size, the age, tumor site and size, ECOG score status or PS score, patients' willingness, lymph node metastasis and the tolerance were all taken into consideration of before deciding surgery. The surgical procedure conclude lumpectomy, total or modi ed radical mastectomy and breast reconstruction. In the rst respective study of NECB based on SEER, the study cohort was limited to large cell neuroendocrine carcinoma and large cell neuroendocrine carcinoma [26]. Their study suggested that surgery improved the DSS of patients. In another similar studies also based on SEER database, surgery was not considered as an independent factors in improving survival [8]. The method that surgery improved DSS and OS was approved by our results. The similar trend was also observed in other case series and small sample respective studies [15].
The adjuvant therapy method were based on the biological feature of NECB such as pathological type, lymph node and distant metastasis [27]. Compare to surgery, chemotherapy was also an essential treatment strategies in improving the survival of NECB [22]. However, the effect of chemotherapy was controversial in previous studies. Some researchers did not consider that patients bene t from chemotherapy. They found that the DSS and OS were decreased because of the drug resistance caused by chemotherapy [28]. Some articles demonstrated that chemotherapy decreased the 12% risk of recurrence in ten years [29]. Our outcomes supported that chemotherapy was signi cant improved the survival of NECB patients. For breast carcinoma, received radiation therapy was an important method in improving survival [30]. The survival analysis showed that radiation was ine cient in prolonging the survival. There was no signi cant difference between the patients received radiation and those not treated with radiation. Similar results were published in previous studies [27].
Besides the traditional treatment strategies, there was speci c regimens in treating neuroendocrine carcinoma, adjuvant endocrine therapy [31,32]. Patients with hormone receptor-positive were candidates to treat with adjuvant endocrine therapy [32][33][34]. Particularly, it was quite suitable for elder,not tolerance surgery or chemotherapy patients [35][36][37]. But some patients rejected to receive endocrine treatment for lacking established procedure or guidelines in NECB. The treatment effect of endocrine therapy was also con ict. It was reported that anti-hormone therapy signi cant prolonged the OS and disease free survival while some other reports did not approved the results [33,38]. The e cient of endocrine therapy need further explore. To triple negative patients, hormone therapy was insu cient. And these patients were correlated with poor prognosis [39,40]. The proportion of triple negative group in NECB patients was larger than IDCB, which is a potential results in explaining relative poor prognosis of NECB.
There were several limitations in our research. Firstly, the detail treatment of patients not provided in the SEER database, making it was impossible to analysis the surgical method, chemotherapy regimens, radiotherapy dose and hormone receptors. It was hardly to analysis the combination treatment effect. Secondly, selection bias may exist in the study for some subgroups just included in limitation sample size patients.

Conclusions
Our study is a large sample, PSM research comparing the demographic and clinical characteristics and risk factors of survival of NECB and IDCB. The analysis showed that NECB patients had advanced SEER stage, more advanced T stage, larger proportion of lymph node metastasis and distant metastasis, and worse OS and DSS compared to IDCB patients. To molecular subtype, NECB tends to be luminal A group. Surgery and chemotherapy, respectively, signi cantly improved the survival of NECB patients.

Declarations Funding No
Con ict of Interest Statement: All authors declared no con ict of interest.
Availability of data and material All the relative data and material were available Authors' Contribution Conception and design: QFL, YWZ and XLM collected and assembly of data; YNZ and QFL do the data analysis and interpretation; QFL and YWZ write the manuscript.
Consent to participate All the author approve Consent for publication All the author approve  Kaplan-Meier curves of (A) DSS and (B) OS in patients