Nomogram prediction for breast cancer-specific survival in patients with ER⁻/PR⁻/HER2⁺ breast cancer: an analysis based on the SEER population

Introduction

According to the estimates of cancer incidence and mortality released by the International Agency for Research on Cancer in GLOBOCAN 2022, breast cancer (BC) is the most common type of cancer among women, with approximately 2.3 million new cases, accounting for 11.6% of all cancers. This disease is the fourth leading cause of cancer deaths globally, with 666,000 deaths (accounting for 6.9% of all cancer deaths) (1). The treatment and prognosis of BC are linked to the expression of human epidermal growth factor receptor 2 (HER2) and hormone receptor (HR), including the estrogen receptor (ER) and progesterone receptor (PR) (2-5). Positive ER and PR suggest that cells can receive signals from estrogen and/or progesterone, and drugs through the endocrine system can inhibit tumor growth (6,7); HER2 is a transmembrane protein of which its intracellular structural region contains sites that bind to adenosine triphosphate, which can reduce the expression of cancer suppressor genes. Its positive expression indicates that cells are growing and proliferating vigorously, which is correlated with tumor invasion and metastasis. Targeted treatment is often recommended for patients with HER2-positive BC (8,9).

Given the critical correlation of ER, PR, and HER2 with treatment and prognosis in BC patients, BC can be classified into four clinical subtypes based on their expression in tumor cells. Triple negative (ER−/PR⁻/HER2⁻), HR-negative and HER2-positive (ER⁻/PR⁻/HER2⁺), HR-positive and HER2-positive, and HR-positive and HER2-negative (10). Based on statistics from the National Cancer Institute and the Centers for Disease Control and Prevention, the American Cancer Society found that from 2017 to 2021, 70% of BC subtypes in the United States were HR⁺/HER2⁻, 9% were HR⁺/HER2⁺, 4% were HR⁻/HER2⁺, and 10% were HR⁻/HER2⁻. Although the HR⁻/HER2⁺ subtype represents the smallest proportion, it has the second lowest 5-year survival rate at 86%, surpassed only by the triple-negative subtype, which has the poorest prognosis at 78% (11). This may be associated with the higher invasiveness and risk of locoregional recurrence of the ER⁻/PR⁻/HER2⁺ subtype, as well as its negative impact on survival following distant metastasis (12-15). Therefore, further in-depth research into this type of BC is still warranted.

In conventional survival analyses, such as the Kaplan-Meier method or the Cox proportional hazards model, cancer-specific mortality (CSM) is considered the outcome of interest, while deaths from other causes are censored. However, this approach ignores the competing risk posed by non-cancer deaths, which can introduce bias when estimating the cumulative incidence of CSM. In clinical practice, BC patients frequently die due to a variety of causes, and these causes compete with one another. The likelihood of CSM may be impacted by non-cancer-specific deaths, and there may be a chance that they will compete with one another. Because conventional survival analysis techniques do not account for the influence of competing risk factors on patient death, they may overestimate the risk of CSM (16). To analyze survival data with competing risks, Austin et al. (17) proposed a method called the competing risks model (CRM), which took into account the effect of other risk factors on CSM and was more consistent with clinical practice to study patient outcomes. CSM from conventional survival analysis was higher than that (18) from CRM.

Due to BC’s biological heterogeneity, the current staging systems based on Roman Numeral and tumor-node-metastasis (TNM), even integrating molecular subtyping information, are not sufficient to fully address clinical needs. In addition, other factors, including marital status, age, adjuvant therapy, and race, may affect prognosis. It has been demonstrated that the presentation of a nomogram as a prediction model outperforms traditional staging systems and shows significant strengths in the prediction of tumor recurrence, prognosis, and outcome (19).

Currently, nomograms have been developed for most BC subtypes, but because the ER⁻/PR⁻/HER2⁺ subtype accounts for the smallest proportion, there have been few studies on this subtype. The current study investigated the survival prognosis and clinical characteristics of patients with ER⁻/PR⁻/HER2⁺ BC via the Surveillance, Epidemiology, and End Results (SEER) database, and established a nomogram to accurately predict the BC-specific survival (BCSS) of patients with ER⁻/PR⁻/HER2⁺ BC based on independent prognostic factors. We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-aw-2562/rc).

Methods

Patient selection and data acquisition

The National Cancer Institute created SEER, a database that gathers cancer patients’ information. The current study utilized SEER*Stat to acquire data on BC patients diagnosed in the period from 2010 to 2015. These cases were analyzed to include female patients with ER⁻/PR⁻/HER2⁺ BC and exclude patients with the following conditions: race, tumor stage, axillary lymph node (LN) positive grade I–II, local LN biopsy, local LN biopsy positive, bone/lung metastasis, tumor size unknown, no surgery or radiotherapy/status of surgery or radiotherapy unknown, survival time <1 month, unmarried or cohabiting, patients with non-primary tumors. Finally, 1,525 eligible subjects were selected for analysis (Figure 1). Because the SEER database’s information is accessible to the general public, the current study did not require patients’ informed consent or ethical approval. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Figure 1 Flowchart of patient selection. AJCC, American Joint Committee on Cancer; ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; LN, lymph node; PR, progesterone receptor; SEER, Surveillance, Epidemiology, and End Results.

Variable collection

The SEER database’s demographic information incorporated year of diagnosis, age recode with single ages and 100+, marital status at diagnosis, and race recode (White, Black, American Indian/Alaska Native, Asian/Pacific Islander). Clinical indicators included primary site-labeled, International Classification of Diseases for Oncology, 3rd Edition (ICD-O-3) histological behavior, grade (through 2017), laterality, combined summary stage (2004+), derived American Joint Committee on Cancer (AJCC) stage group, 7th ed [2010–2015], derived AJCC M, 7th ed [2010–2015], derived AJCC N, 7th ed [2010–2015], derived AJCC T, 7th ed [2010–2015], RX Summ-Surg Prim Site (1998+), RX Summ-Surg/Rad Seq, radiation recode, chemotherapy recode, regional nodes positive (1988+), regional nodes examined (1988+), SEER Combined Mets at DX-lung (2010+), SEER Combined Mets at DX-liver (2010+), SEER Combined Mets at DX-bone (2010+), sequence number, first malignant primary indicator, CS tumor size [2004–2015], Total number of in situ/malignant tumors for patients.

Outcome

The current study selected BC-specific death (BCSD) as the outcome of interest, and deaths from other causes were considered as competing risk events. Survival time is defined as the period from diagnosis to death from BC. To facilitate subsequent study and analysis, the current study renamed the included indicators and assigned values to them (Table 1).

Statistical analysis

R software (4.3.1) was employed for statistical analysis. Frequencies and percentages were utilized to express categorical variables, and the Chi-squared test was conducted to compare differences between groups. Mean ± standard deviation (SD) or interquartile range (IQR) was employed to express continuous variables. The 1,525 patients with ER⁻/PR⁻/HER2⁺ BC were randomized into the training (n=1,067) and validation (n=458) sets in 7:3. Multicollinearity among the candidate variables was assessed using the variance inflation factor (VIF). Variables with a VIF <10 were included in the model to ensure the absence of severe multicollinearity. Significant prognostic factors were initially identified through univariable analysis using the Fine-Gray test. Variables with a P value <0.05 were subsequently entered into a multivariable analysis. Then, a nomogram was constructed using R software to predict 1-, 2-, and 3-year cancer-specific survival (CSS) for patients with ER⁻/PR⁻/HER2⁺ BC. The receiver operating characteristic (ROC) curve and area under the curve (AUC)/concordance index (C-index) were adopted to quantitatively assess the model’s efficacy and predictive power. Calibration curves were utilized to validate the model and to assess the nomogram’s performance. In addition, the performance of CRM and conventional survival analysis was compared regarding 1-, 2-, and 3-year CSS. P<0.05 indicated statistical significance.

Results

Clinicopathologic and demographic characteristics

This study incorporated 1,525 ER⁻/PR⁻/HER2⁺ BC patients and randomized them in 7:3 into a training set (n=1,067) and a validation set (n=458). The demographic and clinical characteristics of all patients are illustrated in Table 2. Most patients had an age of 40 to 70 years, representing 75% of all patients. Among the histologic types, infiltrating duct carcinoma (IDC) was the dominant one, accounting for 94.6% of all patients. Among the grades (through 2017), patients with grade III/IV accounted for the largest proportion (72.9%). Tis/T1 (49.8%), stage N0 (60.7%), and M0 (97.2%) accounted for the largest proportion in the AJCC T/N/M staging systems, respectively. The tumor size was predominant in the 10–20 mm group, making up 28.3% of all patients. The total number of deaths was 203, of which 73 were non-specific, accounting for 36.0% of all deaths (Table 2).

Nomogram based on CRM

Univariable analysis revealed that marital status at diagnosis, age, grade, stage, TNM stage, surgical method, number of LN biopsy, number of positive LN biopsy, tumor size, lung metastasis, and liver metastasis were independent risk factors for specific death in ER⁻/PR⁻/HER2⁺ BC patients. A multivariable analysis containing these factors indicated that age [<40 years as reference, hazard ratio =0.532; 95% confidence interval (CI): 0.284–0.997 for 40–70 years, hazard ratio =0.909; 95% CI: 0.409–2.018 for >70 years], stage (stage I as reference, hazard ratio =2.001; 95% CI: 0.758–5.279 for stage II, hazard ratio =4.938; 95% CI: 1.467–16.629 for stage III/IV), lung metastasis (with metastasis as reference, hazard ratio =0.198; 95% CI: 0.049–0.8 for no metastasis), and tumor size (<10 mm as reference, hazard ratio =1.963; 95% CI: 0.732–5.264 for 10–30 mm, hazard ratio =3.255; 95% CI: 1.015–10.441 for >31 mm) were predictors (Table 3). According to the results, a nomogram based on a CRM was created to predict 1-, 2-, and 3-year CSS for ER⁻/PR⁻/HER2⁺ BC patients (Figure 2).

Figure 2 Nomogram for predicting 1-, 2-, and 3-year breast cancer-specific survival of patients with ER⁻/PR⁻/HER2⁺ breast cancer. ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; PR, progesterone receptor.

Validation of nomogram

The current study employed C-index, ROC curve, and AUC to evaluate the predictive accuracy of nomograms, and their calibration performance was verified by calibration curves. In the training set, the C-index of nomograms was 0.780 (SE: 0.007), and the AUC for 1-, 2-, and 3-year CSS prediction was 0.825 (95% CI: 0.729–0.921), 0.794 (95% CI: 0.716–0.873), and 0.781 (95% CI: 0.724–0.838), respectively (Figure 3), demonstrating a good predictive performance of the model. In the validation set, the C-index was 0.742 (SE: 0.015), and the AUC for 1-, 2-, and 3-year CSS prediction was 0.864 (95% CI: 0.71–1), 0.793 (95% CI: 0.668–0.917), and 0.772 (95% CI: 0.667–0.876), respectively (Figure 3), indicating high predictive value and high reliability of the model in internal validation. The calibration plots indicated good consistency between the nomogram’s prediction and the actual observation in the training and validation sets. Calibration results are illustrated in Figure 4.

Figure 3 ROC curves of the nomogram for the prediction of 1-, 2-, and 3-year CSS in the model training cohort (A) and the validation cohort (B). AUC, area under the curve; CI, confidence interval; CSS, cancer-specific survival; ROC, receiver operating characteristic.

Figure 4 Calibration curves of the nomogram for the prediction of 1-, 2-, and 3-year CSS in the model training cohort (A) and the model validation cohort (B). CSS, cancer-specific survival.

Discussion

In this study, a SEER-based CRM was applied to introduce nomograms to predict CSS in ER⁻/PR⁻/HER2⁺ BC patients at 1, 2, and 3 years, and its performance was evaluated via a validation set. The results indicated that the nomogram exhibited high reliability and significant predictive value.

In the current study, it was found that only 130 of the 1,525 ER⁻/PR⁻/HER2⁺ patients died of BC, and 1,395 survived or died of other diseases. Therefore, CRM was applied to predict prognostic factors for ER⁻/PR⁻/HER2⁺ BC.

Nomograms are a common tool for evaluating oncologic and medical outcomes. Its ability to produce a single numerical probability of a clinical event by integrating various prognostic and determinant variables satisfies the desire for integrated clinical and biological models and promotes the development of personalized medicine (19). The nomogram based on multivariable analysis has good accuracy as a comprehensive statistical model. Previous studies revealed that nomograms had advantages in the evaluation of cancer risks, selection of therapies and drugs, and prediction of survival outcomes for various cancers (19,20).

The current study indicated the important role of age in affecting the prognosis of ER⁻/PR⁻/HER2⁺ BC. Since the age of 40 years is a reasonable cutoff for defining “young” (21,22), this study set three age groups: <40 years, 40–70 years, and >70 years. It was found that BC patients aged >70 years had the highest risk of death. Hershman et al. also discovered that whether the endpoint was overall survival (OS) or BCSD, the prognosis was worse in older BC patients (23). The reason might be that older patients were unable to tolerate intense standard chemotherapy and appeared to be more susceptible to cardiotoxicity from chemotherapy than younger patients (23). Additionally, the incidence of other chronic diseases (e.g., diabetes mellitus) that affected survival was higher in older patients. Nomograms in the current study demonstrated that BC patients aged <40 years exhibited a higher risk of death than those aged 40–70 years. The finding was similar to that of Chen et al. (24), LWingo P et al. (25), and Yancik et al. (26). The reason might be related to the tissue type of BC patients aged <40 years. Azim et al. revealed that regardless of subtype, grade, and stage, the expression of RANK ligand, c-kit, mammary stem and luminal progenitor cells, and BRCA1 mutation signatures were higher in young patients (27). Increasing evidence also suggests that differences in the breast matrix of young patients, as well as changes during pregnancy and lactation, may contribute to differences in tumor biology that emerge later (28).

The prognosis of metastatic breast cancer (MBC) is known to be worse than that of non-metastatic BC. Common sites for distant metastasis of BC include bone, liver, lung, and brain (29). The current study indicated that liver metastasis was most common in patients with ER⁻/PR⁻/HER2⁺ BC, followed by bone metastasis and lung metastasis. Kennecke et al. (14) also found that compared to luminal A tumors, HER2-rich breast tumors had a higher chance of developing liver metastases. Similarly, Leone et al. (30) utilized the SEER database to reveal that liver metastasis was more common in the HR⁻/HER2⁺ subtype. Some studies suggest that CXCR4 is a chemokine receptor that is upregulated by HER2 activation and has been proposed to be involved in promoting tumor cell invasion into the liver (31). However, other studies have reported that liver metastasis is not linked to BC subtypes (32). The current study further analyzed the effect of bone, liver, and lung metastasis on prognosis and indicated that only lung metastasis had an obvious effect on the prognosis of HR⁻/HER2⁺ BC patients. However, Gerratana et al. (29) reported that compared to BC patients with metastases of bone (44.4 months), liver (36.7 months), or brain (7.35 months) as the first distant metastasis, those with lung metastasis exhibited the best survival outcomes (58.5 months). The reason for inconsistent findings might be that Gerratana et al. did not conduct separate studies of HER2-overexpressing BC and the sample size in our study was small.

Tumor size has long been considered a significant prognostic factor for BC, and it has been reviewed by various investigators via data from the Medline database (33), SEER database (34), and Norway Cancer Registry (35). Therefore, the components of the AJCC staging system include tumor size, regional LN status, and distant metastasis (36). The current study revealed that the tumor size of ER⁻/PR⁻/HER2⁺ BC was also an important factor in patient prognosis. When the tumor size was below 10 mm, the risk of death was lowest, while the risk of death was highest when it was above 31 mm. This was comparable to what Rosenberg et al. discovered (34,37). However, Yu et al. found that the correlation of tumor size with breast cancer-specific mortality (BCSM) was segmented in LN-negative diseases. Using tumors of 21 to 30 mm as a reference, the hazard ratio of BCSM rose as tumor size increased until it peaked at a tumor size of 41 to 50 mm, after which the increase in tumor size was unexpectedly correlated with a decrease in the risk ratio, with the lowest point at a tumor size of 61 to 80 mm (38). Furthermore, Wo et al. suggested that small tumors with four LN positives might predict a higher BCSM compared to larger tumors (39).

BC in the SEER database is classified primarily by the differentiation degree of tumor cells. Studies have shown that the classification of BC was important for the prognosis of patients (40,41). The current study illustrated that among all prognostic factors, tumor grade had the greatest impact on the prognosis of HER2-overexpressing BC patients, and the nomogram score of BC patients with poorly differentiated and undifferentiated tumor grades (III/IV) reached 100. Poorly differentiated or undifferentiated BC cells have the characteristics of obvious karyotype mutations and structural disorders. From a biological point of view, such cells tend to have stronger invasiveness and metastasis, and may also have stronger immune escape capabilities. Therefore, such tumors are more malignant and have strong drug resistance and reduced sensitivity to chemotherapy and radiotherapy, greatly affecting patient prognosis.

However, there were limitations to the current study. Firstly, it was a retrospective study, and selection bias was inevitable. Secondly, due to the limitations of public databases, it is impossible to fully study the impact of endocrine therapy, specific radiotherapy and chemotherapy regimens, lifestyle, and family history on the outcomes. For instance, the application of pertuzumab from 2010 to 2015 might have affected the prognosis of patients. Thirdly, our model is designed specifically for treatment-naïve, de novo ER⁻/PR⁻/HER2⁺ BC. It may not be applicable to patients with recurrent or metastatic disease. Finally, as no external validation set was included, the application of nomograms to risk groups in Chinese patient cohorts was not validated. To increase the precision and integrity of nomograms, prospective data can be collected subsequently, and more prospective validation can be included in future studies.

Conclusions

A prediction model was successfully established using the SEER database to accurately predict CSS in patients with ER⁻/PR⁻/HER2⁺ BC based on independent prognostic factors. This model exhibited good performance in both the training and validation sets. It can assist medical professionals in creating customized treatment and follow-up plans.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-aw-2562/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-aw-2562/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-aw-2562/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The 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. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Siegel RL, Miller KD, Fuchs HE, et al. Cancer Statistics, 2021. CA Cancer J Clin 2021;71:7-33. [Crossref] [PubMed]
2. Bartlett JM, Brookes CL, Robson T, et al. Estrogen receptor and progesterone receptor as predictive biomarkers of response to endocrine therapy: a prospectively powered pathology study in the Tamoxifen and Exemestane Adjuvant Multinational trial. J Clin Oncol 2011;29:1531-8. [Crossref] [PubMed]
3. Kurozumi S, Matsumoto H, Hayashi Y, et al. Power of PgR expression as a prognostic factor for ER-positive/HER2-negative breast cancer patients at intermediate risk classified by the Ki67 labeling index. BMC Cancer 2017;17:354. [Crossref] [PubMed]
4. Slamon DJ, Clark GM, Wong SG, et al. Human breast cancer: correlation of relapse and survival with amplification of the HER-2/neu oncogene. Science 1987;235:177-82. [Crossref] [PubMed]
5. Press MF, Bernstein L, Thomas PA, et al. HER-2/neu gene amplification characterized by fluorescence in situ hybridization: poor prognosis in node-negative breast carcinomas. J Clin Oncol 1997;15:2894-904. [Crossref] [PubMed]
6. Jensen EV, Jordan VC. The estrogen receptor: a model for molecular medicine. Clin Cancer Res 2003;9:1980-9.
7. Ueno T. Biomarkers of neoadjuvant/adjuvant endocrine therapy for ER-positive/HER2-negative breast cancer. Chin Clin Oncol 2020;9:35. [Crossref] [PubMed]
8. Gianni L, Eiermann W, Semiglazov V, et al. Neoadjuvant chemotherapy with trastuzumab followed by adjuvant trastuzumab versus neoadjuvant chemotherapy alone, in patients with HER2-positive locally advanced breast cancer (the NOAH trial): a randomised controlled superiority trial with a parallel HER2-negative cohort. Lancet 2010;375:377-84. [Crossref] [PubMed]
9. Ross JS, Slodkowska EA, Symmans WF, et al. The HER-2 receptor and breast cancer: ten years of targeted anti-HER-2 therapy and personalized medicine. Oncologist 2009;14:320-68. [Crossref] [PubMed]
10. Curigliano G, Burstein HJ, Winer EP, et al. De-escalating and escalating treatments for early-stage breast cancer: the St. Gallen International Expert Consensus Conference on the Primary Therapy of Early Breast Cancer 2017. Ann Oncol 2017;28:1700-12.
11. Giaquinto AN, Sung H, Newman LA, et al. Breast cancer statistics 2024. CA Cancer J Clin 2024;74:477-95. [Crossref] [PubMed]
12. Voduc KD, Cheang MC, Tyldesley S, et al. Breast cancer subtypes and the risk of local and regional relapse. J Clin Oncol 2010;28:1684-91. [Crossref] [PubMed]
13. Lobbezoo DJ, van Kampen RJ, Voogd AC, et al. Prognosis of metastatic breast cancer subtypes: the hormone receptor/HER2-positive subtype is associated with the most favorable outcome. Breast Cancer Res Treat 2013;141:507-14. [Crossref] [PubMed]
14. Kennecke H, Yerushalmi R, Woods R, et al. Metastatic behavior of breast cancer subtypes. J Clin Oncol 2010;28:3271-7. [Crossref] [PubMed]
15. Kast K, Link T, Friedrich K, et al. Impact of breast cancer subtypes and patterns of metastasis on outcome. Breast Cancer Res Treat 2015;150:621-9. [Crossref] [PubMed]
16. Meng X, Hao F, Ju Z, et al. Conditional survival nomogram predicting real-time prognosis of locally advanced breast cancer: Analysis of population-based cohort with external validation. Front Public Health 2022;10:953992. [Crossref] [PubMed]
17. Austin PC, Lee DS, Fine JP. Introduction to the Analysis of Survival Data in the Presence of Competing Risks. Circulation 2016;133:601-9. [Crossref] [PubMed]
18. Li Z, Shi Y, Wu L, et al. Establishment and verification of a nomogram to predict tumor-specific mortality risk in triple-negative breast cancer: a competing risk model based on the SEER cohort study. Gland Surg 2022;11:1961-75. [Crossref] [PubMed]
19. Balachandran VP, Gonen M, Smith JJ, et al. Nomograms in oncology: more than meets the eye. Lancet Oncol 2015;16:e173-80. [Crossref] [PubMed]
20. Iasonos A, Schrag D, Raj GV, et al. How to build and interpret a nomogram for cancer prognosis. J Clin Oncol 2008;26:1364-70. [Crossref] [PubMed]
21. Fu J, Wu L, Fu W, et al. How Young Is Too Young in Breast Cancer?-Young Breast Cancer Is Not a Unique Biological Subtype. Clin Breast Cancer 2018;18:e25-39. [Crossref] [PubMed]
22. Zheng H, Ge C, Lin H, et al. The impact of age on outcomes of breast cancer in different hormone receptor and HER2 groups. PLoS One 2023;18:e0280474. [Crossref] [PubMed]
23. Hershman DL, Till C, Shen S, et al. Association of Cardiovascular Risk Factors With Cardiac Events and Survival Outcomes Among Patients With Breast Cancer Enrolled in SWOG Clinical Trials. J Clin Oncol 2018;36:2710-7. [Crossref] [PubMed]
24. Chen HL, Zhou MQ, Tian W, et al. Effect of Age on Breast Cancer Patient Prognoses: A Population-Based Study Using the SEER 18 Database. PLoS One 2016;11:e0165409. [Crossref] [PubMed]
25. Wingo PA, Ries LA, Parker SL, et al. Long-term cancer patient survival in the United States. Cancer Epidemiol Biomarkers Prev 1998;7:271-82.
26. Yancik R, Wesley MN, Ries LA, et al. Effect of age and comorbidity in postmenopausal breast cancer patients aged 55 years and older. JAMA 2001;285:885-92. [Crossref] [PubMed]
27. Azim HA Jr, Partridge AH. Biology of breast cancer in young women. Breast Cancer Res 2014;16:427. [Crossref] [PubMed]
28. Azim HA Jr, Brohée S, Peccatori FA, et al. Biology of breast cancer during pregnancy using genomic profiling. Endocr Relat Cancer 2014;21:545-54. [Crossref] [PubMed]
29. Gerratana L, Fanotto V, Bonotto M, et al. Pattern of metastasis and outcome in patients with breast cancer. Clin Exp Metastasis 2015;32:125-33. [Crossref] [PubMed]
30. Leone JP, Leone BA, Zwenger AO, et al. The prognostic significance of metastatic pattern in stage IV male breast cancer at initial diagnosis: a population-based study. Breast Cancer Res Treat 2021;187:237-44. [Crossref] [PubMed]
31. Li YM, Pan Y, Wei Y, et al. Upregulation of CXCR4 is essential for HER2-mediated tumor metastasis. Cancer Cell 2004;6:459-69. [Crossref] [PubMed]
32. Park HS, Kim S, Kim K, et al. Pattern of distant recurrence according to the molecular subtypes in Korean women with breast cancer. World J Surg Oncol 2012;10:4. [Crossref] [PubMed]
33. Mirza AN, Mirza NQ, Vlastos G, et al. Prognostic factors in node-negative breast cancer: a review of studies with sample size more than 200 and follow-up more than 5 years. Ann Surg 2002;235:10-26. [Crossref] [PubMed]
34. Zheng YZ, Wang L, Hu X, et al. Effect of tumor size on breast cancer-specific survival stratified by joint hormone receptor status in a SEER population-based study. Oncotarget 2015;6:22985-95. [Crossref] [PubMed]
35. Johansson ALV, Trewin CB, Fredriksson I, et al. In modern times, how important are breast cancer stage, grade and receptor subtype for survival: a population-based cohort study. Breast Cancer Res 2021;23:17. [Crossref] [PubMed]
36. Boffa DJ, Greene FL. Reacting to changes in staging designations in the 7th edition of the AJCC staging manual. Ann Surg Oncol 2011;18:1-3.
37. Rosenberg J, Chia YL, Plevritis S. The effect of age, race, tumor size, tumor grade, and disease stage on invasive ductal breast cancer survival in the U.S. SEER database. Breast Cancer Res Treat 2005;89:47-54.
38. Yu KD, Jiang YZ, Chen S, et al. Effect of large tumor size on cancer-specific mortality in node-negative breast cancer. Mayo Clin Proc 2012;87:1171-80. [Crossref] [PubMed]
39. Wo JY, Chen K, Neville BA, et al. Effect of very small tumor size on cancer-specific mortality in node-positive breast cancer. J Clin Oncol 2011;29:2619-27. [Crossref] [PubMed]
40. Zhong X, Lin Y, Zhang W, et al. Predicting diagnosis and survival of bone metastasis in breast cancer using machine learning. Sci Rep 2023;13:18301. [Crossref] [PubMed]
41. Huang X, Luo Z, Liang W, et al. Survival Nomogram for Young Breast Cancer Patients Based on the SEER Database and an External Validation Cohort. Ann Surg Oncol 2022;29:5772-81. [Crossref] [PubMed]