Nontherapeutic Risk Factors of Different Grouped Stage IIIC Breast Cancer Patients' Mortality: A Study of the US Surveillance, Epidemiology, and End Results Database

Objectives Stage IIIC breast cancer, as a local advanced breast cancer, has a poor prognosis compared with that of early breast cancer. We further investigated the risk factors of mortality in stage IIIC primary breast cancer patients and their predictive value. Methods We extracted data from the US Surveillance, Epidemiology, and End Results (SEER) database of female patients with stage IIIC primary breast cancer (n = 1673) from January 2011 to December 2015. Results Hormone receptor negativity (P ≤ 0.001 and P ≤ 0.001, respectively), aggressive molecular typing (P ≤ 0.001 and P ≤ 0.001, respectively), high T stage (P ≤ 0.001 and P ≤ 0.001, respectively), a high number of positive lymph nodes (≥14) (P=0.005 and P=0.001, respectively), and lymph node ratio (≥0.8148) (P ≤ 0.001 and P ≤ 0.001, respectively) were associated with poor disease-specific survival. The indicators of disease-specific survival included estrogen receptor status, progesterone receptor status, molecular typing, T stage, number of positive lymph nodes, and lymph node ratio (P ≤ 0.001,P ≤ 0.001,P ≤ 0.001,P ≤ 0.001, P=0.002, and P ≤ 0.001, respectively). Conclusion Hormone receptor negativity, aggressive molecular typing, high T stage, high number of positive lymph nodes, and lymph node ratio are poor prognostic factors patients with stage IIIC primary breast cancer. The efficient indicators of disease-specific survival include estrogen receptor status, progesterone receptor status, molecular typing, T stage, number of positive lymph nodes, and lymph node ratio.


Introduction
In 2020, the number of new cases of breast cancer worldwide reached 2.26 million, surpassing lung cancer to rank the first in the world [1]. e prognosis of breast cancer is relatively good, and the 5-year survival rate of some breast cancer can reach as high as 99% [2]. However, studies on the staging system showed that the survival rate of breast cancer decreased with the increasing stage [3]. e 7th version of the American Joint Committee on Cancer (AJCC) TNM staging system [4] defines stage IIIC (any TN3M0) breast cancer as N3 stage breast cancer without any distant metastasis. Patients who meet one of the following conditions are diagnosed as pN3: 10 or more axillary lymph nodes metastases; subclavian lymph nodes metastases; clinically found internal mammary lymph nodes metastases with one or more axillary lymph nodes metastases; more than 3 axillary lymph nodes metastases, accompanied by clinically undetected internal mammary lymph nodes metastases confirmed by sentinel lymph node biopsy; and ipsilateral supraclavicular lymph nodes metastases. According to the aforementioned criteria, patients can be further divided into N3a, N3b, and N3c. e 5-year overall survival (OS) and 5-year diseasespecific survival (DSS) of stage IIIC breast cancer are about 61.7% and 66.8%, respectively, and its prognosis is relatively worse than that of early breast cancer [5]. Some studies have discussed the prognostic indicators of stage IIIC breast cancer, such as molecular typing and lymph node ratio (LNR). However, most of the patients were from a single institution and the factors involved were not comprehensive; so, they could not be further grouped [6][7][8]. Based on the US Surveillance, Epidemiology, and End Results (SEER) database, we studied the risk factors of breast cancer-specific mortality (BCSM) in patients with stage IIIC breast cancer and explored their predictive value at different levels.

Patients.
e data supporting this study were obtained from the SEER registry. Since 1975, the SEER program has continuously collected data on cancer incidence and mortality in the United States, which is based on 9 population-based registries. We acquired the required data from the SEER database through SEER * Stat software of version 8.3.9.2.
In the client-server mode of SEER * Stat, we used the International Classification of Diseases for Oncology, the 3rd edition of histology, and behavior codes (ICD-O-3 morphology codes 8500-8599) to identify patients. We selected 2211 female patients diagnosed with stage IIIC primary breast cancer from January 2011 to December 2015, excluding those with unclear pathological classification (n � 146), missing immunohistochemical markers (n � 69), unknown involved lymph node number (n � 215), inaccessible lymph node staging (n � 77), and incomplete follow-up (n � 31). Finally, a total of 1673 patients were left ( Figure 1). . e measurement data were described by median (interquartile range). Frequency was used to show the counting data. e Kaplan-Meier method was used in survival analysis, and the log-rank test was used to compare different survival curves. Univariate and multivariate analyses were performed using the Cox model. Finally, the ROC analysis was conducted to illustrate the discriminating ability of risk factors in terms of BCSM and report the area under the curve (AUC). Statistical significance was set as two-sided P < 0.05, and all confidence intervals (CI) were expressed at the 95% confidence level.

Clinicopathological Features.
Most patients with stage IIIC breast cancer were over 40 years of age (91.15% vs 8.85%). As for the histological type, patients with invasive ductal carcinoma predominated (64.02% vs 35.98%). In terms of molecular typing, luminal A patients ranked first, followed by triple negative patients (68.02% and 12.91%, respectively). e number of ER positive patients, PR positive patients, and HER-2 negative patients were far higher than their counterparts (78.00%, 65.03%, and 80.93%, respectively). e proportions of T2 stage patients and N3a stage patients were the highest (45.13% and 85.24%, respectively). ere were more patients with an LNP higher than 14 and patients with an LNR higher than 0.8148 (56.37% and 55.29%) ( Table 1). e median follow-up time was 54 (38∼72) months. At the end of the follow-up, 578 patients died, of whom 447 died of breast cancer. e 5-year OS and DSS were 51.72% (95% CI: 0.4838∼0.5495) and 60.74% (95% CI: 0.5735∼0.6395), respectively.

Multivariate Analysis by the Cox Model Assessing Factors
Associated with DSS. In multivariate analysis, ER negativity, PR negativity, aggressive molecular typing, and high T stage were still associated with poor DSS (P ≤ 0.001,P ≤ 0.001, P ≤ 0.001, and P ≤ 0.001, respectively). A higher number of LNP (≥14) and LNR (≥0.8148) were also associated with poor DSS (P � 0.005 and P ≤ 0.001, respectively). ere was a borderline level of statistical significance between the N3c stage and DSS (Hazard ratio: 1.419, 95% CI: 0.982∼2.051,P � 0.063) ( Table 3).    Figure 2(a). e survival rate of patients with hormone receptor (HR) positivity was significantly higher (P ≤ 0.001 and P ≤ 0.001, respectively; Figures 2(b) and 2(c)). e prognosis of patients with an aggressive molecular typing was poor (P ≤ 0.001, Figure 2(d)). Moreover, the higher the T stage and N3 stage were, the lower the survival rate of patients was (P ≤ 0.001 and P � 0.0004, respectively; Figures 2(e) and 2(f )). We also found that patients with LNP <14 and LNR <0.8148 had a better prognosis (P � 0.0008 and P ≤ 0.001, respectively; Figures 2(g) and 2(h)).

Predictive Value of Risk Factors.
e ROC curve and AUC showed the discrimination ability of the ER status, PR status, T stage, molecular typing, LNP number, and LNR in stage IIIC breast cancer (Figure 3). e indicators for predicting DSS in patients with stage IIIC breast cancer included ER status, PR status, molecular typing, Tstage, LNP number, and LNR (P ≤ 0.001, P ≤ 0.001, P ≤ 0.001, P ≤ 0.001, P � 0.002, and P ≤ 0.001, respectively; Table 4). At the same time, we reported the ROC curve and AUC of the abovementioned predictors in patients with different N3 stages (Table 5 and Figure 4). e results showed that the predictors of DSS in patients with N3a stage were ER status, PR status, molecular typing, T stage, LNP number, and LNR (P ≤ 0.001, P ≤ 0.001, P ≤ 0.001, P ≤ 0.001, P ≤ 0.001, and P ≤ 0.001, respectively). e indicators with a good predictive value for DSS in patients with N3b stage only included the ER status, PR status, molecular typing, and T stage (P � 0.020, P � 0.024, P � 0.009, and P � 0.007, respectively); however, these indicators had a poor predictive ability for DSS in N3c patients. Meanwhile, there was no significant difference in the DSS prediction ability of each index in patients with different N3 stages (P > 0.05).

Discussion
e 5-year OS of stage IIIC breast cancer was about 61%∼ 73%, and it was reported that the prognosis was not good [9,10]. e median follow-up time of patients in this study was 54 (38∼72) months, the restricted mean DSS was 76.027 (95% CI: 74.5258∼77.5287) months, and the 5-year OS and 5-year DSS were 51.72% (95% CI: 0.4838∼0.5495) and 60.74% (95% CI: 0.5735∼6395), respectively. However, not all patients with stage IIIC breast cancer have the same clinicopathological characteristics and prognosis. For example, some researchers believed that the risk of BCSM in HR-negative stage IIIC breast cancer patients in 20 years was significantly higher than that in HR-positive patients (68.7% vs 63.1%) [11].
At the beginning, we discussed the clinicopathological features of patients with stage IIIC breast cancer and found that stage IIIC breast cancer patients had old age, low HER-2 positivity rate, and a high HR positivity rate. In terms of molecular typing, Luminal A patients ranked first, followed by triple negative patients (68.02% and 12.91%, respectively). e HR and HER-2 status are recognized prognostic markers in breast cancer. Compared with other subtypes, the overall prognosis of triple negative breast cancer was poor [12,13], while the overall prognosis of Luminal B breast cancer was the best [14,15]. Our Kaplan-Meier analysis also confirmed this point. e survival rates of HR-positive patients were higher than those of HR-negative patients (P ≤ 0.001). As for the histological type, patients with invasive ductal carcinoma predominated (64.02% vs 35.98%). After matching, invasive ductal carcinoma had similar OS to invasive lobular carcinoma, but invasive lobular carcinoma patients with risk factors had worse outcomes [16]. In addition, a high proportion of LNP <14 and LNR <0.8148 were also the characteristics of these patients. Notably, after breast cancer patients with supraclavicular nodal metastases downstaged from AJCC stage IV to IIIC, N3c patients who received the standard therapy demonstrated better OS than stage IV disease [17,18]. Hence, we sought to illustrate whether there was a difference between stage IIIC patients.
In this study, univariate analysis showed that poor molecular typing, ER negativity, PR negativity, HER-2 negativity, high T stage, and high N3 stage were associated with poor DSS (P ≤ 0.001, P ≤ 0.001, P ≤ 0.001, P � 0.017, P ≤ 0.001 and P ≤ 0.001, respectively). In addition, the prognosis of patients with stage IIIC breast cancer varied with different N3 stages. Real-world studies found that patients with the N3c stage had the worst prognosis, followed by N3a patients, while N3b patients had the best prognosis. Both univariate and multivariate analyses showed that HR negativity (P ≤ 0.001 and P ≤ 0.001, respectively), aggressive molecular typing (P ≤ 0.001 and P ≤ 0.001, respectively), high T stage (P ≤ 0.001 and P ≤ 0.001, respectively), a high number of LNP (≥14) (P � 0.001 and P � 0.005, respectively), and LNR (≥0.8148) (P ≤ 0.001, and P ≤ 0.001, respectively) were associated with poor DSS. When the 7th version of the AJCC TNM staging system took N staging and the number of LNP as lymph node-related prognostic indicators, some researchers had questioned whether it was the most accurate way to represent the state of metastatic cancer in lymph nodes and proposed that LNR should be considered to be more valuable [19,20]. In patients with stage II to III breast cancer, LNR may be a more useful tool for predicting survival than pathological lymph node staging [21][22][23].
We drew the ROC curve of the abovementioned prediction indicators and reported the AUC. e predictors of DSS for patients with stage IIIC breast cancer included ER status, PR status, molecular typing, T stage, LNP number, and LNR (P ≤ 0.001, P ≤ 0.001, P ≤ 0.001, P ≤ 0.001, P � 0.002, and P ≤ 0.001, respectively). In some studies, large tumor size, lymph node metastasis, and HR negativity were often considered to be adverse prognostic factors [24][25][26]. In multivariate analysis, older age at diagnosis, higher N stage, and ER  6 e Breast Journal negativity were found to be independent predictors of BCSM [27]. We wanted to know whether the predictive value of these indicators was different from patients with different N3 stages. erefore, patients were further grouped according to the N3 stage and we found that the indexes with predictive value for DSS of N3a patients were ER status, PR status, molecular typing, T stage, LNP number, and LNR (P ≤ 0.001, P ≤ 0.001, P ≤ 0.001, P ≤ 0.001, and P ≤ 0.001, respectively). e Breast Journal e indicators with a good predictive value for DSS in patients with the N3b stage only included ER status, PR status molecular typing, and T stage (P � 0.020 , P � 0.024, P � 0.009, and P � 0.007, respectively). However, these indicators had a poor predictive ability in N3c patients. In previous studies, grade III disease, perineural invasion, LNP ≥20, and LNR ≥0.9 were found to be associated with the OS and disease-free survival of N3a stage breast cancer [28,29]. For N3b stage breast cancer, pathological prognostic staging was an independent predictor related to OS and DSS [30]. In addition, there was no significant difference in the discrimination ability of DSS among patients with different N3 stages in our study (P > 0.05).
Inevitably, our research also has some limitations. For example, due to the lack of detailed follow-up data, we have not discussed the effect of different treatment options on the survival of patients with stage IIIC breast cancer under modern treatment modalities. In addition, we failed to establish an effective prognosis prediction model. We still lack specific exploration and discussion on the predictive value of LNR.

Conclusion
e survival rate of HR-positive patients with stage IIIC breast cancer is significantly higher. e higher the T stage and N3 stage are, the lower the survival rate is. e prognosis of patients with aggressive molecular typing, LNP ≥14, and LNR ≥0.8148 is poor. HR negativity, poor molecular typing, high T stage and N3 stage, and high LNP and LNR are the adverse prognostic factors of stage IIIC breast cancer. e ideal predictors of BCSM in patients with stage IIIC breast cancer include ER status, PR status, molecular typing, T stage, LNP number, and LNR. Although the discrimination ability of the abovementioned indicators in patients with different N3 stages is dissimilar, there is no statistical difference.

Data Availability
All data generated or analyzed during this study are included in this article.
Ethical Approval e authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. e research study was carried out in accordance with the Declaration of Helsinki (as revised in 2013).

Conflicts of Interest
e authors declare that they have no conflicts of interest regarding the publication of this article.

Authors' Contributions
Yue Qiu and Yuhui Chen are the guarantors of the integrity of the study and study concepts and design.
Literature research and clinical studies were conducted by all authors. All authors participated in the experimental studies, data analysis, and statistical analysis. All authors contributed to manuscript preparation and manuscript editing. All authors gave final approval of the version to be submitted.