Primary Tumor Location is Associated with Prognosis for Women with Breast Cancer

Background: The prognostic impact of tumor location on breast cancer patients is not consistent and still controversial. We aimed to investigate the prognostic role of primary tumor location on the survival of patients with breast cancer. Methods: Using the Surveillance, Epidemiology, and End Results database, we identied 53,905 patients diagnosed with tumors in the lower quadrants (n=11,065), upper quadrants (n=38,974), or central and nipple (n=3,866). Chi-squared test was used to compare categorical variables across the groups. Cox proportional hazard models were applied to estimate the factors associated with prognosis. Results: Compared with the other quadrants, patients with central and nipple lesions showed generally more unfavorable clinicopathologic features and worse breast cancer-specic survival (BCSS) and overall survival (OS). Multivariate Cox analysis showed a higher hazard ratio (HR) for tumor location of central and nipple (BCSS: HR, 1.145, p = 0.036, 95% condence interval [CI], 1.009-1.299; OS: HR, 1.118, p = 0.024, 95% CI: 1.015-1.232), while lower HR were observed for upper quadrants (BCSS: HR, 0.888, p = 0.004, 95% CI: 0.818-0.964; OS: HR, 0.930, p = 0.023, 95% CI: 0.873-0.990). Multivariate logistic regression indicated that tumors located in central and nipple were more likely to be inoperable disease (HR, 1.460, 95% CI: 1.300-1.640, p < 0.001), while tumors located in upper quadrants tend to be operable disease (HR, 0.895, p = 0.005, 95% CI, 0.829-0.967). Conclusion: Tumors located in central and nipple had negative contact with BCSS and OS, while tumor located in upper quadrants had favorable contact with survival.


Introduction
Breast cancer is currently the rst most common malignancy in the USA and the second leading cause of cancer mortality in women [1]. Up to now, a number of clinicopathological factors, including hormone receptor (HR) status, human epidermal growth factor receptor-2 (HER2) expression, Ki-67 level, tumor size, pathological grade, and lymph node (LN) status, have been identi ed as powerful indicators predicting prognosis and guiding adjuvant treatments [2][3][4][5]. Furthermore, multigenic molecular tests (i.e. Oncotype DX, MammaPrint, EndoPredict, and Prosigna)) have been developed and facilitated risk strati cation of tumors into low-risk or high-risk groups, which helped predict recurrence and guide chemotherapy [6,3,7]. Despite of all these remarkable prognostic factors, patients with the same subtype are still bearing different survival outcomes. It is essential to nd more effective parameters to help tailor personalized therapy for each individual.
Previous studies have occasionally reported the in uence of primary tumor location on the prognosis of breast cancer, but the results are not consistent and still controversial. Tumors most frequently located in the upper outer quadrant (UOQ). Sohn et al. [8] reported that patients with tumors in the UOQ were associated with an independent contribution to better survival than that of patients with tumors in other quadrants. Some other studies have reported that patients with medial tumors showed lower survival rates than those with lateral tumors [9,10]. Meanwhile, patients with non-UOQ tumors especially lower inner quadrant (LIQ) have been demonstrated with worse survival [11,12]. Moreover, tumors in central portion have been veri ed to have a worse prognosis in some reports [13,14]. Furthermore, Kroman et al. [15] found the risk of dying increased signi cantly (up to 21%) with increasing distance of tumor location from the axilla.
Despite of all these studies, some other authors have reported no correlation between tumor location and breast cancer survival [16,17].
Although upper inner quadrant (UIQ) is also one of inner quadrants like LIQ, its prognostic in uence was quite different from that of LIQ.
Axillary lymph-node metastases (ALNM) is less frequent observed in the UIQ (20.6%) compared with all other quadrants (33.2%) [18]. Hwang et al. [19] analyzed 63,388 patients with primary breast cancer from the Korean Breast Cancer Registry and found the survival rate of UIQ was not inferior to those of UOQ or lower outer quadrant (LOQ). Levi et al. [20] included 4,562 patients and concluded compared to tumors located in UOQ, the HR was 1.02 for UIQ, 1.20 for LOQ, and 1.55 for LIQ. Sarp et al. [12] identi ed 1,522 women with breast cancer recorded at the population-based Geneva Cancer Registry and demonstrated the 10-year speci c survival was 94% for patients with tumors in UOQ, 96% for the UIQ, 91% for the LOQ, the 88% for the LIQ. All these studies have veri ed the different prognostic in uence of tumors in LIQ, and Kamakura et al. [21] elucidated the recurrence-free survival rate was lower in patients with the lower quadrantss carcinoma than other regions. Based on above studies, we rst reclassi ed primary tumor sites into 3 groups such as: lower quadrants (including LIQ and LOQ), upper quadrants (including UIQ and UOQ), and central and nipple. To describe prognostic value of tumor location for patients with breast cancer, we performed a population-based, nationally representative cohort study using the National Cancer Institute's SEER database.

Page 3/11
As an authoritative 18 population-based cancer registries, the SEER program collects and publishes cancer incidence and survival data, which covers approximately 28% of the US population (http://seer.cancer.gov/). Subjects had to meet the following criteria for inclusion: (1) age at diagnosis between 18 and 75 years old; (2) histological type was limited to in ltrating duct carcinoma (International Classi cation of Diseases for Oncology, third revision, ICD-O-3, 8500); (3) known tumor size and lymph node status; (4) surgically resected with pathology specimen and con rmed by pathological diagnosis; (5) known survival time and cause of death; (6) known HR and HER2 status; (7) known surgery type; (8)

Statistical analysis
Cases of yearly diagnosed breast cancer patients were extracted via the SEER*Stat software (version 8.3.6). Baseline characteristics by tumor location were compared using chi-square tests. Breast cancer-speci c survival (BCSS) was calculated from the date of diagnosis to the date of breast cancer-speci c death, and overall survival (OS) was calculated from the time of diagnosis to the time of death by any cause. Kaplan-Meier estimates of BCSS and OS were plotted by tumor location and analyzed across groups with log-rank tests. In addition, risk factors associated with prognosis were analyzed using Cox proportional hazard models. Logistic regression model was utilized to present the relationship between the tumor location and inoperable disease. All computed p values were two-sided, and a p value of less than 0.05 was considered statistically signi cant. All analyses were performed using the R 2.15.3 software.

Baseline characteristics of patients strati ed by tumor location
An initial analysis of the demographic and clinical characteristics of all 53,905 patients were summarized in Table 1. The median follow-up time was 61 months. The most common tumor location was the upper quadrants (72.3%), followed by tumors located in lower quadrants (20.5%), and the central and nipple region (7.2%). Compared with the other quadrants, patients with central and nipple lesions showed generally more unfavorable clinicopathologic features (older age, higher proportion of patients with stage II/III/IV, more patients with T2/T3/T4, and more patients with positive LN). Patients with central and nipple lesions were observed with less lumpetomy surgery rate and a signi cantly lower incidence of HR-/HER2-(triple-negative) subtype (all p < 0.001). Comparison Of Survival Among The Three Groups Using the lower quadrants population as a reference, we found the 5-year BCSS of patients with central and nipple lesions was signi cantly lower than that of the lower quadrants population (90.5% vs 93.2%, p < 0.001), while patients of the upper quadrants showed signi cantly better BCSS (93.8% vs 93.2%, p = 0.01) (Fig. 1A). Similarly, the 5-year OS of the patients with tumors in central and nipple were markedly lower than that of the lower quadrants population (84.5% vs 88.9%, p < 0.001), while patients with tumors in the upper quadrants showed signi cantly better OS (89.8% vs 88.9%, p = 0.004) (Fig. 1B) Fig. 1A, Supplementary Fig. 2A), and triple-negative cancer (BCSS: p = 0.04; OS: p = 0.056) ( Supplementary Fig. 1D, Supplementary Fig. 2D).

Tumors Location Is Related To Tumor Resectability
To clarify the reasons leading to poor prognosis of tumors in central and nipple, we analyzed the relationship between tumor location and inoperable disease of breast cancer (TNM stage III/IV, except for T3N1M0). Variables that were signi cant (p < 0.001) in the univariate logistic analysis (age, race, grade, tumor size, breast subtype, marital status, and tumor location) were further included in the multivariate logistic regression analysis. Tumors located in central and nipple were more likely to be inoperable diseases (HR, 1.460, p < 0.001, 95% CI, 1.300-1.640), while tumors located in the upper quadrants tend to be operable diseases (HR, 0.895, p = 0.005, 95% CI, 0.829-0.967) ( Table 3).

Discussion
Studies on tumor location of different subsets of breast cancer data, however, have produced some inconsistent results. Some studies showed breast cancer patients with tumors located in medial, lower, or central regions may indicate adverse prognostic outcomes [19,11,12,20,14,10,21]. Whereas other studies didn't show association between tumor location and breast cancer prognosis [16,17]. In light of all of these evidences, we hypothesize this lack of consensus and disparate outcomes is partially owing to lack of standardized division of the breast. In our study, we reviewed prior researches and rst introduced triple-classi cation of tumor location as upper quadrants, lower quadrants, and central and nipple. Our classi cation method seemed to be more reasonable, which helped to lead to more moderate results.
In the current study, we found patients with tumors in central and nipple exhibited more unfavorable clinicopathologic characteristics (older age, higher proportion of patients with stage II/III/IV, more patients with T2/T3/T4, and more patients with positive LN), and these clinicopathologic features could partially explain the worse prognosis of tumors in central and nipple. This may because X-rays can hardly penetrate in the nipple-areolar complex, as a result, tumors in central and nipple were harder to be detected and may reach a substantially larger size when mammography could screen out. Ansari et al. [22] reported that breast cancer located closer to the nipple have a higher incidence of metastases to axillary LN, which is in accordance with our study. Our nding is similar to Gill's study [14], which concluded that the HR of central tumors for overall survival was 1.46 mostly due to their more advanced stages. Besides, Ji et al. also demonstrated tumors in the central and nipple portion is an independent adverse factor for BCSS and OS [13].
Compared with lower quadrants, we elucidated the favorable prognostic value of upper quadrants in various subgroups of breast cancer patients. Accordingly, pertinent studies have reported better prognosis of patients with tumors in UOQ [23,8], and tumors in LIQ have been veri ed as worse prognostic factor in breast cancer [24,12]. Anatomically, the lymphatics from the lower portion of the breast appear to occasionally communicate with those under the diaphragmatic region or peritoneal plexus by passing through the sheath of rectus abdominis, and Kamakura and colleagues [21] revealed that a lower quadrants tumor location was a signi cant prognostic factor for recurrence, especially soft tissue and visceral recurrence. Furthermore, we hypothesized that the worse prognosis of tumors in lower quadrants is partially because of possible hidden internal mammary node (IMN). Vendrell-Torné et al. [25] found that drainage from the LIQ in 30% of cases occurred exclusively to IMN. In particular, Shahar et al. [26] recently reported that drainage to the internal mammary chain was signi cantly more seem in lower tumors comparing upper tumors (lower 36.4% vs upper 14.6%, p = 0.003). Adjuvant chemotherapy has become a standardized therapy to reduce mortality in LN positive patients [27], while adjuvant radiotherapy of the internal mammary chain is still controversial [28]. Hwang et al. [19] revealed LIQ tumors showed worse OS in LN negative patients who received no chemotherapy, but similar OS was seen in patients who received chemotherapy. Because of possible hidden IMN metastasis, lower tumors could be understaged and under-treated, leading to worse survival outcomes.
Inevitably, there are some limitations in the present study. First, our study is retrospective in nature, further validations from other institutions are merited. Secondly, we had no access to details of chemotherapy regimens, radiation or other adjuvant therapies. Thus, no de nitive statements can be made with respect to whether the trends we presented are caused by treatment differences. Despite of these shortcomings, we rst introduced this triple-classi cation and obtained remarkable results, it contributes to the growing literature regarding prognostic value of tumor location on breast cancer.
In summary, we have provided evidence that the tumor location of central and nipple was an independent adverse prognostic factor, while the tumor location of upper quadrants was an independent favorable prognostic factor. Further studies are impending to verify whether tumor location could help guide more personalized treatment algorithms.

Declarations Funding
No.

Con icts of Interest
The authors have no conflicts of interest to disclose.
Availability of data and material Not applicable.
Code availability Not applicable.
Author's contributions Zhuowei Tang, Yuzhu Ji: Study design, data collections, statistical analysis, writing and manuscript preparation.
Ethics approval Not applicable.

Consent to participate
Not applicable.

Consent for publication
The authors agree for publication.