Development and validation of a SEER-derived prognostic nomogram for non-metastatic Asian breast cancer patients

Introduction

Breast cancer (BC) ranks as the most common cancer among American women, with projections indicating that 30% of new female cancer diagnoses in 2024 will be for BC (1,2). While the Asian American population has grown by 81% since 2000, significantly increasing the absolute burden of disease, the age-adjusted incidence rate of BC in this group has also been rising steadily. Recent data from 2024 indicate an annual increase of approximately 2.5% to 2.7% in Asian American/Pacific Islander women, a trend that is outpacing other racial groups (3-5). Additionally, Asian Americans often confront unique challenges within the healthcare system, exacerbated by cultural, social, health literacy, and language barriers, which further escalate disparities in cancer outcomes (6-9). Thus, understanding survival outcomes and relevant risk factors for Asian American BC patients is vital (10).

The conventional tumor-node-metastasis (TNM) staging system is extensively utilized for prognostic analysis; however, significant prognostic disparities persist among patients classified under the same TNM stage, especially as treatment modalities evolve (6,11-13). Recently, nomograms have emerged as innovative predictive tools, complementing the TNM system by incorporating an expanded range of prognostic factors (14-16). Prior studies have identified critical determinants of BC survival, including age, tumor size, grade, and various treatment modalities such as surgery, chemotherapy, and radiotherapy (4,17-19). Despite numerous predictive models based on the Surveillance, Epidemiology, and End Results (SEER) database, no study has specifically sought to create a prognostic nomogram tailored for the Asian American BC population.

This study seeks to fill this gap by investigating relevant factors influencing BC survival among Asian American patients through data from the SEER database, ultimately providing a reference for clinicians to accurately assess prognostic factors in this demographic. We present this article in accordance with the TRIPOD+AI reporting checklist (available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0203/rc).

Methods

Data source and data extraction

Data regarding Asian female patients diagnosed with BC between 2000 and 2021 were acquired from the SEER database, adhering strictly to the established research guidelines. Collected clinicopathological data included age, marital status, grade, TNM stage, income, surgical history, radiotherapy, chemotherapy, subtype, survival status, and survival time. Data retrieval was conducted utilizing SEER*Stat software (version 8.4.4). Inclusion criteria encompassed: (I) confirmed pathological diagnosis of BC; (II) primary tumor; (III) accurate clinical information. Exclusion criteria included: (I) non-Asian population; (II) distant metastasis at diagnosis; (III) incomplete clinicopathological data, particularly unknown HER2 or hormone receptor status; (IV) survival times of less than one month. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Statistical analysis

Statistical analysis and modeling were conducted using R 4.1.2. Missing data were handled via complete case analysis. Patients with missing values for key clinicopathological variables (molecular subtype, TNM stage, tumor grade, marital status, or treatment information) were excluded during study enrollment, as these variables were considered essential for reliable prognostic modeling. The proportions of excluded patients due to missing data are detailed in the patient flow diagram (Figure 1). No imputation method was applied in this study. All predictors were categorized according to standard clinical classification conventions to ensure clinical interpretability. Age was dichotomized at 75 years, a widely used threshold for defining geriatric BC populations with distinct prognostic and treatment characteristics. Tumor grade, TNM stage, and molecular subtype were classified according to the American Joint Committee on Cancer (AJCC) staging system and standard pathological classification criteria. Marital status, radiotherapy, and chemotherapy were categorized as defined in the SEER database coding manual. The “caret” package facilitated the random division of patients into training and validation groups at a ratio of 7:3. Intergroup comparisons of categorical data employed the Pearson chi-squared test. Overall survival (OS) was defined as the time from diagnosis until death for any reason. As a secondary outcome, cancer-specific survival (CSS) was defined as the time from diagnosis to death attributed to BC, with deaths from other causes treated as censored for CSS analysis. Competing risk analysis using the Fine-Gray model was also performed for CSS, treating non-breast-cancer mortality as a competing event. As a sensitivity analysis, LASSO penalized Cox regression was performed to validate the variable selection process. Variables with P<0.05 from univariate Cox regression were subjected to least absolute shrinkage and selection operator (LASSO) regression to eliminate redundant predictors, and the screened variables were further fitted into multivariate Cox regression for subsequent multivariate adjustment. The selected variables contributed to the construction of the nomogram, with its accuracy determined through calibration curves that compared predicted OS to actual OS rates. A 10-fold cross-validation approach was implemented to circumvent overfitting, and bootstrap resampling was carried out 1,000 times to minimize overfitting bias. The discriminative ability of the nomogram was assessed using the Concordance Index (C-index), while its clinical utility was evaluated through decision curve analysis (DCA). Time-dependent AUC values at 1, 3, 5, and 10 years were calculated to evaluate time-specific discriminative performance. Brier scores were computed to assess overall prediction accuracy. Net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were calculated to compare the incremental prognostic value of the nomogram versus the traditional TNM staging system.

Figure 1 Flow diagram of patient selection. BCS, breast-conserving surgery; SEER, Surveillance, Epidemiology, and End Results.

Results

Patient characteristics

Overall, 38,269 patients met the inclusion criteria, comprising 26,789 in the training group and 11,480 in the validation group (Figure 1). The training dataset was utilized for model development and evaluation, while the validation dataset served to reassess model performance. Validation results were only accepted when baseline characteristics were comparable between the training and validation sets. The baseline clinicopathological characteristics are summarized in Table 1, which shows no significant differences between the two groups (P>0.05).

Selection of independent prognostic factors

Univariate and multivariate Cox regression analyses performed on the training set revealed the identified variables meeting a significance threshold of P<0.05. LASSO penalized regression identified the same independent prognostic factors, confirming the robustness of the variable selection. Independent factors such as age, grade, marital status, T stage, N stage, subtype, radiotherapy, and chemotherapy were validated in the multivariate analysis, providing essential predictors of OS (Figure 2). The full specifications of the multivariate Cox regression model, including hazard ratios, and 95% confidence intervals (CIs) for each predictor, are provided in Figure 2.

Figure 2 Multivariate Cox regression analyses of overall survival in the training cohort. ***, P<0.001. CI, confidence interval; DSW, divorced, separated, widowed; HR, hormone receptor; HER2, human epidermal growth factor receptor 2.

Construction and validation of nomogram

Based on the independent indicators found in the multivariate analysis, a novel prognostic nomogram was constructed for predicting individual survival at 1-, 3-, 5-, and 10-year intervals (Figure 3). Each patient’s score was calculated by summing the scores from each prognostic factor, enabling the estimation of the probabilities of 1-, 3-, 5-, and 10-year OS.

Figure 3 Nomogram of the current prognostic model for individualized survival predictions. N, node; OS, overall survival; T, tumor.

Assessment of predictive performance of the nomogram

The prognostic model exhibited strong discriminative capabilities, with C-indices of 0.794 (95% CI: 0.782–0.805) and 0.811 (95% CI: 0.795–0.827) for training and validation cohorts, respectively. Bootstrap resampling (1,000 repetitions) yielded an optimism-corrected C-index of 0.796 in the training cohort, indicating minimal overfitting. These results surpassed those of the traditional TNM staging system, which recorded C-indices of 0.730 (95% CI: 0.716–0.745) and 0.740 (95% CI: 0.718–0.762). Time-dependent AUC values for OS were 0.789 (1-year), 0.792 (3-year), 0.790 (5-year), and 0.785 (10-year) in the training cohort, with consistent results in the validation cohort. The Brier scores at 1, 3, 5, and 10 years were 0.012, 0.048, 0.087, and 0.162 in the training cohort, indicating good predictive accuracy. Compared with the TNM staging system, the nomogram yielded a continuous NRI of 0.312 (95% CI: 0.278–0.345) and an IDI of 0.041 (95% CI: 0.035–0.047) for 5-year OS, demonstrating significant improvement in risk reclassification. Calibration plots for the 1-, 3-, 5-, and 10-year OS indicated a strong correlation between the predicted and observed outcomes in both training and validation cohorts (Figure 4). Furthermore, DCA highlighted the nomogram’s considerable clinical utility in predicting OS rates at 1, 3, 5, and 10 years (Figure 4). The X-axis represents the threshold probability at which a clinician would classify a patient as high-risk and consider intensified treatment or surveillance. The net benefit of the nomogram exceeds that of the TNM system across the clinically relevant threshold range of 5% to 30%, which corresponds to typical decision thresholds for adjuvant therapy in early BC. For CSS, the nomogram achieved C-indices of 0.782 (95% CI: 0.769–0.795) in the training cohort and 0.795 (95% CI: 0.777–0.813) in the validation cohort. Competing risk analysis using the Fine-Gray model yielded consistent prognostic effects for all predictors, confirming the stability of the results.

Figure 4 Assessment of predictive performance of the prognostic model. (A) Calibration plot of the nomogram model at 1-, 3-, 5- and 10-year of OS in the training cohort. (B) Calibration plot of the nomogram model at 1-, 3-, 5- and 10-year of OS in the validation cohort. DCA curves comparing the net benefit of the nomogram and traditional TNM staging for overall survival prediction across a range of risk threshold probabilities in the training (C) and validation (D) cohorts. AUC, area under the curve; DCA, decision curve analysis; OS, overall survival; TNM, tumor-node-metastasis.

Discussion

This study is pioneering in developing and validating a prognostic nomogram for predicting the long-term survival of Asian BC patients using data from the SEER database. The nomogram integrates vital independent prognostic factors established through comprehensive multivariate modeling, including patient age, tumor grade, marital status, T stage, N stage, subtype, and treatment modalities such as radiotherapy and chemotherapy. This intuitive predictive tool demonstrated robust discriminative and calibration performance, exceeding that of the traditional TNM staging system in both training and validation cohorts. The improvement in C-index compared with TNM staging (~0.06 absolute difference) represents a modest but clinically meaningful gain, enabling more refined risk stratification beyond anatomical staging alone. This magnitude of improvement is consistent with other published nomograms for BC.

Several prognostic nomograms for BC have been developed using the SEER database in recent years (14-16). However, most were constructed in mixed-race cohorts with White patients as the majority, and Asian patients were only included as a small subgroup for exploratory analysis rather than as the target population for model development. Our model is specifically trained and validated in a large, exclusively non-metastatic Asian American cohort, which may provide more accurate survival estimates for this population than general-purpose nomograms. We acknowledge that our model relies on conventional clinicopathological factors and does not include genomic biomarkers, detailed treatment regimens, or granular socioeconomic metrics, which are not captured in the SEER database. Therefore, the present work provides an incremental improvement in prognostic tools for Asian American BC patients, rather than a paradigm-shifting advance.

The identification of these clinicopathological characteristics as independent predictors of OS among Asian BC patients is consistent with existing literature (20-23). Older age at diagnosis has a well-documented correlation with poorer outcomes, attributed to increased treatment-related toxicities, competing comorbidities, and intrinsic tumor biological differences prevalent in older patients (24,25). Furthermore, higher tumor grades, advanced nodal involvement, and unfavorable subtypes (e.g., triple-negative) are established adverse prognostic factors across racial and ethnic groups (26,27). Our findings reinforce the clinical relevance of these variables for risk stratification specific to the Asian BC population.

Marital status emerged as an independent prognostic factor in this cohort, with unmarried and divorced/separated/widowed patients showing worse survival outcomes than married patients (28,29). Marital status should be interpreted as a proxy marker for social support, healthcare access, and other social determinants of health, rather than a causal factor. This finding aligns with previous studies demonstrating associations between social support and cancer outcomes, potentially mediated by better treatment adherence, reduced psychological distress, and improved access to care among married individuals (30-36). Including this sociodemographic factor in our predictive model underscores the multifaceted nature of cancer prognosis beyond clinical and tumor characteristics.

Moreover, the provision of multimodal cancer therapies—radiotherapy and chemotherapy—independently influenced survival within our Asian BC cohort (11,37-41). While further investigation is necessary to explore the biological interactions between these therapies and tumor characteristics, our results stress the need for comprehensive treatment details during prognostic model development. This nomogram is designed as a practical supplementary tool to support clinical decision-making for non-metastatic Asian American BC patients. Clinicians can calculate a patient’s total score by summing points for each clinicopathological factor, and use the corresponding survival probabilities for personalized risk stratification and patient counseling. Based on the distribution of total scores in our cohort, patients can be stratified into risk groups: low risk (total score <15.65 points), and high risk (≥15.65 points). For high-risk patients, the model can help identify individuals who may benefit from intensified adjuvant systemic therapy, radiotherapy, or more frequent post-treatment surveillance. For low-risk patients, the tool may support shared decisions to de-escalate treatment to avoid unnecessary toxicity. Notably, this nomogram is intended to complement, not replace, clinical judgment and multidisciplinary team assessment.

Several limitations should be acknowledged. First, this study only performed internal validation via random split-sample and bootstrap resampling, and no external validation using independent cohorts from different healthcare systems or regions was conducted. This represents a major limitation, and the generalizability of the nomogram to non-US Asian populations or other healthcare settings remains to be confirmed; accordingly, the model is not yet ready for widespread global clinical application. Second, the retrospective nature of this analysis inherently invites selection bias. Additionally, the SEER database lacks specific clinical variables, including performance status, comorbidities, and detailed systemic therapy information. Future prospective studies encompassing a more extensive array of predictors, as well as external validation in independent multi-institutional Asian BC cohorts across diverse healthcare settings, are warranted to further enhance the nomogram’s accuracy and generalizability.

Conclusions

In conclusion, this study presents a novel prognostic nomogram that effectively predicts long-term survival for Asian BC patients, outperforming traditional TNM staging. By incorporating readily available clinicopathological factors, this user-friendly tool provides a foundation for personalized risk assessment, though further external validation is required before widespread clinical implementation.

Acknowledgments

We would like to thank the patients who participated in this study.

Footnote

Reporting Checklist: The authors have completed the TRIPOD+AI reporting checklist. Available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0203/rc

Peer Review File: Available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0203/prf

Funding: This work was supported by the grants from the National Natural Science Foundation of China (project number 82503927), the Guangdong Basic and Applied Basic Research Foundation (project number 2026A1515012552), the Fundamental Research Funds for the Central Universities (project number YG2024QNA06), and the Supporting Research Fund for High-level Full-time Talents Introduction of Guangdong Provincial People's Hospital (project number KY012025021).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://gs.amegroups.com/article/view/10.21037/gs-2026-0203/coif). X.H. reports that this work was supported by the grants from the National Natural Science Foundation of China (project number 82503927), the Guangdong Basic and Applied Basic Research Foundation (project number 2026A1515012552), the Fundamental Research Funds for the Central Universities (project number YG2024QNA06), and the Supporting Research Fund for High-level Full-time Talents Introduction of Guangdong Provincial People's Hospital (project number KY012025021). The other 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. Truong A, McKinley M, Gomez SL, et al. The role of ethnic enclaves and neighborhood socioeconomic status in invasive breast cancer incidence rates among Asian American, Native Hawaiian, and Pacific Islander females in California. Cancer Causes Control 2025;36:183-9. [Crossref] [PubMed]
2. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
3. Lei S, Zheng R, Zhang S, et al. Global patterns of breast cancer incidence and mortality: A population-based cancer registry data analysis from 2000 to 2020. Cancer Commun (Lond) 2021;41:1183-94. [Crossref] [PubMed]
4. Giaquinto AN, Sung H, Newman LA, et al. Breast cancer statistics 2024. CA Cancer J Clin 2024;74:477-95. [Crossref] [PubMed]
5. Satagopan JM, Stroup A, Kinney AY, et al. Breast cancer among Asian Indian and Pakistani Americans: A surveillance, epidemiology and end results-based study. Int J Cancer 2021;148:1598-607. [Crossref] [PubMed]
6. Ko NY, Hong S, Winn RA, et al. Association of Insurance Status and Racial Disparities With the Detection of Early-Stage Breast Cancer. JAMA Oncol 2020;6:385-92. [Crossref] [PubMed]
7. Medina HN, Callahan KE, Morris CR, et al. Cancer Mortality Disparities among Asian American and Native Hawaiian/Pacific Islander Populations in California. Cancer Epidemiol Biomarkers Prev 2021;30:1387-96. [Crossref] [PubMed]
8. Eden CM, Johnson J, Syrnioti G, et al. The Landmark Series: The Breast Cancer Burden of the Asian American Population and the Need for Disaggregated Data. Ann Surg Oncol 2023;30:2121-7. [Crossref] [PubMed]
9. Ihenacho U, McKinley MA, Vu A, et al. Characterizing breast cancer incidence and trends among Asian American, Native Hawaiian, and non-Hispanic White women in Hawai'i, 1990-2014. Cancer Causes Control 2023;34:241-9. [Crossref] [PubMed]
10. Liu P, Li X, Mittendorf EA, et al. Comparison of clinicopathologic features and survival in young American women aged 18-39 years in different ethnic groups with breast cancer. Br J Cancer 2013;109:1302-9. [Crossref] [PubMed]
11. Moran MS, Ho AY. Radiation Therapy for Low-Risk Breast Cancer: Whole, Partial, or None? J Clin Oncol 2022;40:4166-72. [Crossref] [PubMed]
12. Swaminathan V, Spiliopoulos MK, Audisio RA. Choices in surgery for older women with breast cancer. Breast Care (Basel) 2012;7:445-51. [Crossref] [PubMed]
13. Han T, Shi M, Chen Q, et al. Effect of adjuvant radiotherapy after breast-conserving surgery in elder women with early-stage breast cancer: a propensity-score matching analysis. Front Oncol 2023;13:1012139. [Crossref] [PubMed]
14. 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]
15. Antonini M, Pannain GD, Mattar A, et al. Systematic Review of Nomograms Used for Predicting Pathological Complete Response in Early Breast Cancer. Curr Oncol 2023;30:9168-80. [Crossref] [PubMed]
16. Wang Z, Xing L, Luo X, et al. A nomogram for survival prediction in 275,812 U.S. patients with breast cancer: a population-based cohort study based on the SEER database. Gland Surg 2022;11:1166-79.
17. Trinh QD, Li H, Meyer CP, et al. Determinants of cancer screening in Asian-Americans. Cancer Causes Control 2016;27:989-98. [Crossref] [PubMed]
18. Wang S. Design and implementation of breast cancer prediction system based on machine learning. Theoretical and Natural Science 2024;32:238-47.
19. Gandhi S, Nie J, Trevisan M, et al. Social networks, social determinants, and mortality: Western New York Exposures and Breast Cancer study. JNCI Cancer Spectr 2024;8:pkae057. [Crossref] [PubMed]
20. Kim EK, Park SY, Kim SW. Clinicopathological characteristics of BRCA-associated breast cancer in Asian patients. J Pathol Transl Med 2020;54:265-75. [Crossref] [PubMed]
21. Chen DN, Song CG, Ouyang QW, et al. Differences in breast cancer characteristics and outcomes between Caucasian and Chinese women in the US. Oncotarget 2015;6:12774-82. [Crossref] [PubMed]
22. Lin MQ, Li YP, Wu SG, et al. Differences in esophageal cancer characteristics and survival between Chinese and Caucasian patients in the SEER database. Onco Targets Ther 2016;9:6435-44. [Crossref] [PubMed]
23. Pineda MD, White E, Kristal AR, et al. Asian breast cancer survival in the US: a comparison between Asian immigrants, US-born Asian Americans and Caucasians. Int J Epidemiol 2001;30:976-82. [Crossref] [PubMed]
24. Schonberg MA, Marcantonio ER, Li D, et al. Breast cancer among the oldest old: tumor characteristics, treatment choices, and survival. J Clin Oncol 2010;28:2038-45. [Crossref] [PubMed]
25. Di Lascio S, Tognazzo E, Bigiotti S, et al. Breast cancer in the oldest old (≥ 89 years): Tumor characteristics, treatment choices, clinical outcomes and literature review. Eur J Surg Oncol 2021;47:796-803. [Crossref] [PubMed]
26. Nolan E, Lindeman GJ, Visvader JE. Deciphering breast cancer: from biology to the clinic. Cell 2023;186:1708-28. [Crossref] [PubMed]
27. Tang Z, Ji Y, Min Y, et al. Prognostic Factors and Models for Elderly (≥70 Years Old) Primary Operable Triple-Negative Breast Cancer: Analysis From the National Cancer Database. Front Endocrinol (Lausanne) 2022;13:856268. [Crossref] [PubMed]
28. Jiao D, Ma Y, Zhu J, et al. Impact of Marital Status on Prognosis of Patients With Invasive Breast Cancer: A Population-Based Study Using SEER Database. Front Oncol 2022;12:913929. [Crossref] [PubMed]
29. Martínez ME, Unkart JT, Tao L, et al. Prognostic significance of marital status in breast cancer survival: A population-based study. PLoS One 2017;12:e0175515. [Crossref] [PubMed]
30. Paranjpe A, Zheng C, Chagpar AB. Disparities in Breast Cancer Screening Between Caucasian and Asian American Women. J Surg Res 2022;277:110-5. [Crossref] [PubMed]
31. Osborne C, Ostir GV, Du X, et al. The influence of marital status on the stage at diagnosis, treatment, and survival of older women with breast cancer. Breast Cancer Res Treat 2005;93:41-7. [Crossref] [PubMed]
32. Ding W, Ruan G, Lin Y, et al. Dynamic changes in marital status and survival in women with breast cancer: a population-based study. Sci Rep 2021;11:5421. [Crossref] [PubMed]
33. Parise C, Caggiano V. The influence of marital status and race/ethnicity on risk of mortality for triple negative breast cancer. PLoS One 2018;13:e0196134. [Crossref] [PubMed]
34. Taparra K, Qu V, Pollom E. Disparities in Survival and Comorbidity Burden Between Asian and Native Hawaiian and Other Pacific Islander Patients With Cancer. JAMA Netw Open 2022;5:e2226327. [Crossref] [PubMed]
35. Freeman JQ, Li JL, Fisher SG, et al. Declination of Treatment, Racial and Ethnic Disparity, and Overall Survival in US Patients With Breast Cancer. JAMA Netw Open 2024;7:e249449. [Crossref] [PubMed]
36. Zhai Z, Zhang F, Zheng Y, et al. Effects of marital status on breast cancer survival by age, race, and hormone receptor status: A population-based Study. Cancer Med 2019;8:4906-17. [Crossref] [PubMed]
37. Walker GA, Kaidar-Person O, Kuten A, et al. Radiotherapy as sole adjuvant treatment for older patients with low-risk breast cancer. Breast 2012;21:629-34. [Crossref] [PubMed]
38. Xie W, Cao M, Zhong Z, et al. Survival outcomes for breast conserving surgery versus mastectomy among elderly women with breast cancer. Breast Cancer Res Treat 2022;196:67-74. [Crossref] [PubMed]
39. Morgan JL, George J, Holmes G, et al. Breast cancer surgery in older women: outcomes of the Bridging Age Gap in Breast Cancer study. Br J Surg 2020;107:1468-79. [Crossref] [PubMed]
40. Okonogi N, Kono S, Karasawa K, et al. Significance of Hypofractionated Radiotherapy in Postoperative Irradiation for Breast Cancer: An Asian Multi-institutional Prospective Study. Clin Oncol (R Coll Radiol) 2023;35:463-71. [Crossref] [PubMed]
41. Lim LY, Miao H, Lim JS, et al. Outcome after neoadjuvant chemotherapy in Asian breast cancer patients. Cancer Med 2017;6:173-85. [Crossref] [PubMed]