Nomogram predicted overall survival of patients with neuroendocrine carcinoma of the breast

Background For primary neuroendocrine carcinoma of breast was a very rare subtype in breast cancers, its prognosis was still controversial and there was no independent standard for its treatment. The purpose of our retrospective study was to construct a nomogram to predict the overall survival (OS) of patients with neuroendocrine carcinoma of the breast. Methods 150 patients of training cohort were collected from Surveillance, Epidemiology, and End Results (SEER) database diagnosed between 2003 and 2015, and 93 patients of verification cohort were enrolled from Sun Yat-sen University Cancer Center (Guangzhou, China) diagnosed between 2004 and 2018. The nomogram was constructed uniting three significantly risk factors of overall survival identified by univariate and multivariate analysis and then validated using receiver operating characteristic (ROC) curves for discrimination, calibration plots and the decision curves analysis (DCA). Results Age, N stage and PR status were closely and significantly related to overall survival in patients with breast neuroendocrine carcinoma. The C-index of nomogram in the training and verification cohorts are 0.775 (95% CI, 0.784 to 0.615) and 0.760 (95% CI, 0.705 to 0.800) respectively. Calibration plots of practical and predicted possibility for the nomogram demonstrated that the predictive 5-year overall survival rate was in accordance with the actual overall survival probability in both sets. Moreover, the decision curves (DCA) also expressed pretty clinical benefit of the nomogram across a range of high-risk threshold. Conclusion This novel population-based nomogram may help with treatment decisions in patients with neuroendocrine carcinoma of the breast (NEBC).

Breast cancer was the number one multifactorial cancer in women worldwide,composed of many different subtypes with various clinicopathologic features. Neuroendocrine carcinoma of the breast is a rare special subtype, endorsed by the WHO for the first time in 2003 [1]. Neuroendocrine tumors of the breast (NEBC) are very uncommon malignant tumors having a proportion of less than one percent in breast cancers [2]. Exactly as a result of its scarce, the existing learning of NECB is confined to clinical case reports and some small series analysis [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18]. And most of the current knowledge concerning the disease was obtained from these studies. For specific recommendations regarding treatment do not exist so far, surgical management, similar to conventional breast cancer, was performed with NEBC patients [19]. Although curative surgical treatment has been performed, the long-term postoperative prognosis of NECB is still controversial.
On the prognosis, no consensus has been reached for NEBC. Contradictory research results appeared in the previous studies, very likely thanks to the limited reported number of patients, and different selection criteria [20]. Thus, identification of prognostic factors of NEBC has long been receiving much attention. And to get to know more about mammary NEC, we have made full use of a relatively complete and large cancer database collected over the past two decades from Surveillance, Epidemiology, and End Results (SEER) registries and Sun Yat-sen University Cancer Center (Guangzhou, China). Using these data, we assessed the onset and clinical prognosis of NEBC.
There were three goals of our present study. First, our aim is to assess the prognostic value of clinicopathological characteristics in NEBC patients. Secondly, we designed to establish a predictive nomogram which has been wildly used in many other cancer diseases to accomplish overall survival estimation for NEBC patients [21][22][23][24][25][26][27]. Finally, we aimed to evaluate the clinical applicability of the nomogram in predicting prognosis.

Patients
Two independent groups of NEBC patients after radically surgical resection were selected for this study. The training cohort patients (n = 150) were from Surveillance,  Table 1. Univariate analysis was carried out to evaluate the statistical significance of clinical and pathological features. Multivariate analysis results were obtained by Cox proportional hazard regression. All the analyzed variables with P values less than 0.05 in the multivariate analysis were used to create a predictive nomogram.
The nomogram was established in the training set and then verified in the other validation set. The discrimination was quantified by the area under the receiver operating characteristic (ROC) curves (AUC or C-index). Graphically evaluation of calibration was executed by using the Hosmer fitting optimization test to draw the relationship between actually observed probabilities and predictive probabilities [28]. The Decision Curve Analysis (DCA) was applied to assess the clinical effectiveness of nomogram for the prognosis prediction. All tests were performed in both sets of queues and P < 0.05 was considered significant.

Clinicopathological characteristics
The comprehensive characteristics of all enrolled patients in the training and verification cohorts were shown in Table 1

Construction and verification of prognostic nomogram for OS
All the three independently predictive factors of OS in multivariate cox proportional hazards regression analysis derived from the training cohort were integrated into the prediction nomogram model which is shown in Fig. 1. The nomogram demonstrated good accuracy with an unadjusted concordance index (C-index) of 0.775 (95% CI, 0.784 to 0.615) ( Fig. 2A). The calibration curve of survival probability for 5 years after operation also showed pretty good consistency between the nomogram model and actual observation prediction (Fig. 2C).
In the verification cohort, the median follow-up time was 30 months (range, 3-169 months). The C-index of the constructed nomogram for predicting 5-year OS was 0.760 (95% CI, 0.705 to 0.800) (Fig. 2B), and good calibration was observed for predicting the probability of 5-year OS (Fig. 2D).
Clinical usefulness of the nomogram as evaluated by DCA Given that the established nomogram demonstrated good predictive capabilities in terms of C-index, DCA was needed to determine the clinical validity of the nomogram. On DCA, the nomogram showed great net benefit with a wide range of threshold probability (Fig. 3A). Then we performed DCA in the verification cohort which also presents a similar wide range of threshold probability (Fig. 3B). This result further demonstrated fairly good estimation of clinical decision making at high net benefit levels.

Discussion
Mammary NEC has always been a controversial disease. Different studies have conveyed different clinical results owing to inconsistent diagnostic criteria in part, Some have revealed that the prognosis of neuroendocrine carcinomas was better than that of patients without specific types of aggressive cancer, whereas a worse prognosis has recently been observed in more larger series [9,14,16,29]. To study more about prognosis of NEBC, we collected total 243 NEBC patients who underwent curative surgery both from the SEER and Most population-based results showed that NECB is often positive for hormone receptors, whereas HER2 is almost always negative. In agreement with these previous series of study, our results also showed that the positive rate of ER, PR and HER2 are 83.5%, 72.4% and 2.9% respectively. Meanwhile, the study also presented that NEBC tended to occur in elderly patients (mean age 59 years), showing a higher pathological grade, a larger tumor (mean 27 mm) and less local lymph node metastasis.
The advantage of our present study was that two cohorts of NEBC patients were enrolled for analysis, and they all underwent radically surgical treatment.

Patient consent for publication
Not applicable.

Availability of data and materials
The datasets used in our study are available from the author on reasonable request.

Competing interests
No potential conflict of interest.

Funding
The study was supported by funds from the National Natural Science Foundation of China (81302318 to Xinhua Xie; 81672598 to Xiaoming Xie), the Science and Technology Planning Projects of Guangdong (2017A030313554 to Weige Tan; 2017A020215163 to Xinhua Xie).

Authors' contributions
WT and XX designed the study and performed statistical analysis of the data. XX, YK and PL conducted the data analysis and interpretation. WT wrote the manuscript. Xiao. X and Wei. Tan edited and revised the manuscript. All authors discussed the results. Five-year overall survival nomogram for patients with NEBC. In order to use the nomogram, the value of each single patient is situated on each variable axis, and a line is drawn up to calculate the number of points received for every variable value. To use the nomogram, the value of an individual patient is located on each variable axis, and a line is drawn up-ward to calculate the number of points received for each variable value. The points were summed up together to obtain the total points and 5-year OS can be estimated the total points of each patient.