Development and internal validation of a multivariable prognostic prediction model in advanced breast cancer patients based on the SEER database

Introduction

According to the American Cancer Society’s 2024 cancer statistics (1), breast cancer will remain the most prevalent malignancy among U.S. women and the projected incidence of new cases is estimated to account for 32% of the total occurrences of female cancers. Concurrently, this disease is expected to cause 42,250 mortalities, underscoring its persistent public health burden despite decades of therapeutic advancements. Based on a 2024 systematic review (2) from the International Agency for Research on Cancer (IARC), approximately 32% of global breast cancer diagnoses occur at advanced stages (III–IV).

Advanced breast cancer (ABC) is defined as encompassing locally advanced breast cancer (LABC) and metastatic breast cancer (MBC) (3). LABC usually includes some stage IIB (T3N0M0) and stage IIIA (T3N1M0) breast cancers that are feasible for radical surgery, and stage IIIB and IIIC breast cancers with skin, chest wall or regional lymph node involvement that are difficult to perform radical surgery (4). MBC includes both recurrence and de novo stage IV disease at initial diagnosis. ABC is widely recognized as a manageable yet refractory to curative intent condition, characterized by its proclivity for recurrence and intrinsic resistance to chemotherapy (5). The aim of treatment in ABC is to prolong life, manage symptoms and improve quality of life (6).

Clinical trials evaluating therapeutic approaches for ABC have reported different findings regarding treatment efficacy. Findings from the Molecular Profiling of Breast Cancer Study (MPBCS) phase III trial demonstrate that breast-conserving surgery (BCS) is associated with significantly improved overall survival (OS) compared to mastectomy in patients with LABC (7). A randomized controlled trial (MF07-01) comparing sequential localized regional therapy (LRT) followed by systemic therapy versus upfront systemic therapy alone in patients with de novo stage IV breast cancer demonstrated that the combined treatment strategy significantly improved 10-year OS compared to monotherapy (8). A retrospective cohort study from the Chinese Academy of Medical Sciences revealed that primary tumor resection in patients with de novo stage IV breast cancer was associated with a statistically significant reduction in cancer-related symptom progression or recurrence (9). According to a multicenter prospective cohort study (TBCRC 013), surgery after chemotherapy for stage IV breast cancer does not improve patients’ survival (10). The prospective phase III trial ABCSG-28 (POSYTIVE) could not demonstrate a benefit of OS for surgical resection of the primary tumors in breast cancer patients presenting with de novo stage IV disease (11). In addition, another randomized controlled trial indicates that primary tumor resection does not significantly improve OS in patients with de novo MBC who achieved clinical response to first-line chemotherapy (12).

Currently, the treatment options for ABC are still controversial. Therefore, this study seeks to systematically characterize the clinicopathological profiles of patients with progressive disease using the Surveillance, Epidemiology, and End Results (SEER) database and develop predictive models and internally validate their predictive performance to guide clinicians in delivering precision medicine through individualized prognostic assessment. We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2588/rc).

Methods

Data source

We used data from the SEER Program (http://seer.cancer.gov/about/overview.html), which was established by the National Cancer Institute (NCI) of the United States. It systematically compiles comprehensive cancer-related data, including demographic information (age, gender, race), tumor characteristics [primary site, histologic type, grade, American Joint Committee on Cancer (AJCC) staging], treatment modalities (surgical resection, radiotherapy, chemotherapy), follow-up outcomes (survival status, cause-specific mortality) and vital status updates. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Study population

We obtained data from the SEER database on a total of 42,857 patients with primary breast cancer diagnosed by pathology from 2010–2015. Specifically, the following demographic information is included: age, race, grade, pathology, T-stage, N-stage, M-stage, surgery, radiation, chemotherapy, metastasis, estrogen receptor (ER) status, progesterone receptor (PR) status, human epidermal growth factor receptor 2 (HER2) status and survival outcomes. Next, we meticulously filtered through the data to identify relevant and meaningful information. The inclusion criteria were as follows: (I) female patients aged ≥15 years at diagnosis; (II) with histopathologically confirmed primary breast cancer; (III) clinically staged as IIB, IIIA, IIIB, IIIC, or TxNxM1 according to the AJCC Cancer Staging Manual (6th Edition). Cases with incomplete information regarding race, grade, ER, PR, HER2 status and distant metastasis status were excluded from the analysis. Eventually, 27,593 patients were enrolled in the study cohort (Figure 1).

Figure 1 Flowchart for screening patient information derived from the SEER database. ABC, advanced breast cancer; ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; LABC, locally advanced breast cancer; PR, progesterone receptor; SEER, Surveillance, Epidemiology, and End Results.

Statistical analysis

We randomized all patients into two groups, training set and validation set. Chi-squared test was performed to assess balance in baseline variables including age, race, grade, and biomarker status (ER/PR/HER2) and treatment modalities (surgery/radiotherapy/chemotherapy), confirming no statistically significant differences (P>0.05). OS is defined as the time from the date of diagnosis to death from any cause, while breast cancer-specific survival (BCSS) refers to the time from the date of diagnosis to breast cancer-related death. Univariate survival analysis using Kaplan-Meier curves and log-rank tests was performed to evaluate the prognostic impact of clinicopathological variables on OS and BCSS. Multivariate Cox proportional hazards regression analysis was conducted to evaluate independent predictors of breast cancer outcomes and statistically significant variables (P<0.05) were incorporated into the nomogram construction. The statistical methods described above were performed using SPSS 27.0 (IBM Corp, Armonk, NY, USA).

Variables achieving statistical significance (P<0.05) in multivariate analysis were selected for inclusion in the nomogram construction. Nomogram performance was validated via decision curve analysis (DCA) to assess clinical utility, receiver operating characteristic (ROC) curves to evaluate discriminative ability and calibration curves to measure prediction accuracy. Area under the curve (AUC) represents the area under the ROC curve, with higher values indicating stronger discriminative power for predicting target events. R version 4.2.0 was used for nomogram construction (rms package), ROC curve analysis (riskRegression and survival packages), calibration curve evaluation (rms and survival packages), and DCA (ggDCA, rms, and survival packages).

Results

Demographic and clinical characteristics

A total of 27,593 ABC cases meeting the inclusion criteria were identified from the SEER database and randomly assigned to a training set (n=19,315, 70.0%) and a validation set (n=8,278, 30.0%) using a randomized allocation strategy (Table S1).

ABC includes LABC and MBC. Categorization was performed based on the 6th edition of the AJCC Cancer Staging Manual, with LABC defined as stages IIB, IIIA, IIIB and IIIC, and MBC as stage M1. In the training dataset, 13,158 cases (68.1%) were classified as the LABC group and 6,157 cases (31.9%) as the MBC group (Table 1). There was no significant difference in disease grading between these two groups, which were dominated by G2 (LABC: 5,368, 40.8%; MBC:2,575,41.8%) and G3 (LABC: 6,519, 49.5%; MBC: 3,079, 50.0%). Furthermore, most patients had lymph node metastases (LABC: 9,405, 71.5%; MBC: 4,889, 79.4%). Bone metastasis predominated as the most common site of distant metastasis, observed in 3,919 patients (20.3%), followed by liver metastasis (1,500 cases, 7.8%), lung metastasis (1,883 cases, 9.7%) and brain metastasis (403 cases, 2.1%). The incidence of invasive ductal carcinoma (IDC) was slightly higher in MBC at 4,747 cases (77.1%) compared to LABC with 8,629 cases (65.6%). In addition, the number of patients choosing surgical intervention was higher in the MBC group (3,612, 58.7%) than in the LABC group (1,417, 10.8%). Notably, MBC patients received radiotherapy and chemotherapy less frequently compared with their LABC counterparts: 2,219 cases (36.0%) vs. 7,403 cases (56.3%) for radiotherapy, and 3,837 cases (62.3%) vs. 9,726 cases (73.9%) for chemotherapy respectively.

In the training dataset, 14,230 patients (73.7%) underwent surgical intervention, while 5,085 patients (26.3%) were not operated. The main clinicopathological features and treatment strategies are summarized in Table 2. Compared to the surgery group, patients in the non-surgery group were older and had more advanced T-stages. Of note, the non-surgery group demonstrated a significantly higher proportion of distant metastases at 3,651 cases (71.8%) relative to the surgical group with 2,506 cases (17.6%). Meanwhile, patients in the surgery cohort underwent more comprehensive treatment regimens, including chemotherapy (75.1% vs. 56.6%, respectively) and radiotherapy (59.4% vs. 23.0%, respectively), compared to the non-surgical cohort. The majority of patients in both groups were hormone receptor positive with comparable rates, whereas IDC prevalence was higher in the non-surgery group. Collectively, the non-surgery group demonstrated more advanced disease status at baseline.

Survival analysis and prognostic risk factors

We evaluated the impact of the following predictors on OS and BCSS: age, race, tumor grade, T-stage, N-stage, M-stage, surgical type, chemotherapy, radiation therapy, ER status, PR status, HER2 status, histological subtype, and sites of distant metastasis (bone, brain, liver, and lung). All variables were analyzed as categorical factors.

In the LABC cohort, 3-year OS and BCSS rates were 79.8% and 83.8%, respectively. Five-year OS and BCSS were 69.2% and 76.1%, while 10-year OS and BCSS were 53.3% and 65.4%. In univariate Kaplan-Meier survival analysis (Table 3), all evaluated variables (i.e., age, race, tumor grade, tumor size, lymph node status, molecular subtype, metastasis, surgery, radiation, chemotherapy and pathology) were statistically significant predictors of both OS and BCSS (P<0.05). However, in multivariate Cox proportional hazards regression analysis (Table 4), no statistically significant association was observed between pathology and either OS or BCSS. In addition, bone metastases did not independently predict BCSS [hazard ratio (HR): 3.629, 95% confidence interval (CI): 0.51–25.89, P=0.20].

In the MBC cohort, 3-year OS and BCSS rates were 48.2% and 51.3%, respectively. Five-year OS and BCSS were 31.7% and 35.4%, while 10-year OS and BCSS were 15.9% and 20.2%. In contrast to the LABC group, in MBC, pathology was not associated with notable differences in OS (P=0.15) and BCSS (P=0.09) in univariate Kaplan-Meier survival analysis (Table 5). Moreover, N-stage failed to independently predict BCSS (P=0.10), and bone metastasis was not associated with OS (P=0.09) outcomes. Multivariate Cox proportional hazards analysis (Table 6) demonstrated no significant survival benefit of radiotherapy on OS (HR: 0.959, 95% CI: 0.901–1.022, P=0.20) or BCSS (HR: 0.991, 95% CI: 0.928–1.059, P=0.80) in MBC patients. In addition, after adjusting for confounding variables (e.g., molecular subtypes and metastatic sites), G4 grade lost its prognostic significance in multivariate analysis, suggesting its lack of independent predictive utility (OS: HR: 1.012, 95% CI: 0.591–1.731, P=0.97; BCSS: HR: 0.978, 95% CI: 0.548–1.747, P=0.94). Patients in the MBC group, who were older, non-White race and had larger tumor size, and brain, liver, or lung metastases, had a worse prognosis, while surgery, chemotherapy, hormone receptor positivity, and HER2 positivity were identified as independent protective factors.

In the surgery group, 3-year OS and BCSS rates were 78.1% and 81.7%, respectively. Five-year OS and BCSS were 67.2% and 73.0%, while 10-year OS and BCSS were 50.5% and 61.1%. Kaplan-Meier survival analysis demonstrated that 17 clinical factors were statistically significant predictors of OS and BCSS (P<0.001) (Table S2). Although statistically significant, mastectomy demonstrated a relatively modest adverse prognostic effect on survival outcomes compared to partial mastectomy in the multivariate model (OS: HR: 1.083, 95% CI: 1.011–1.161, P=0.02; BCSS: HR: 1.111, 95% CI: 1.024–1.205, P=0.01) (Table S3). Notably, brain and liver metastases were identified as the most potent predictors of poor survival outcomes, exerting a significantly greater negative impact compared to other distant metastatic sites.

Construction and validation of nomogram

Using results from multivariate Cox proportional hazards regression, significant independent predictors of OS and BCSS were plotted in a nomogram to predict 3-, 5-, and 10-year survival. Validation of the nomogram’s predictive accuracy included ROC curve analysis for discriminative ability, calibration curves for calibration-in-the-large, and DCA for clinical benefit assessment. The nomogram features a points scale [0–100] at the top. First, the values for each predictor variable axis are ascertained for the patient. Then, these values are mapped vertically upward to the corresponding values on the numeric axis. Subsequently, the individual factor scores are summed to derive the total score. Finally, the total score is then located on the total score axis, from which survival probabilities across specified time points are obtained by projecting vertically downward to the survival probability axis.

In the LABC group, eleven clinicopathological variables (age, race, grade, T-stage, N-stage, surgery, radiation, chemotherapy, ER-status, PR-status, HER2-status) were incorporated into the nomogram development for OS and BCSS predictions (Figure 2A,2B). In the MBC group, twelve prognostic factors were selected, including age, race, grade, T-stage, surgery, chemotherapy, ER-status, PR-status, HER2-status, brain metastasis, liver metastasis and lung metastasis, which were used to construct nomograms for OS and BCSS assessments (Figure 2C,2D). In the surgery group, sixteen variables were included in the nomogram construction. These encompassed age, race, grade, T-stage, N-stage, M-stage, surgery, radiation, chemotherapy, ER-status, PR-status, HER2-status, bone metastasis, brain metastasis, liver metastasis and lung metastasis (Figure 2E,2F).

Figure 2 Nomograms for predicting 3-, 5-, and 10-year OS and BCSS. Predicting survival of OS (A) and BCSS (B) in the LABC group. Predicting survival of OS (C) and BCSS (D) in the MBC group. Predicting survival of OS (E) and BCSS (F) in the surgery group. BCSS, breast cancer-specific survival; ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; LABC, locally advanced breast cancer; MBC, metastatic breast cancer; OS, overall survival.

The clinical predictive performance and credibility of the model were evaluated using ROC curves, with high AUC values demonstrating robust discriminative ability between survivors and non-survivors. Figure 3 and Table 7 illustrate that each group exhibited AUC values exceeding 75%, indicating robust discriminative performance across all evaluated patient subsets. Calibration curves (Figure 4) indicated excellent agreement between predicted 3-, 5-, and 10-year survival probabilities and actual observations, underscoring the model’s clinical reliability. Additional internal validation was performed via bootstrap resampling (≥1,000 resamples) (Figure S1). The DCA curve (Figure 5) demonstrated that the predictive model offered good clinical utility, helping clinicians to choose a model with a higher net benefit in the target threshold range.

Figure 3 ROC curves evaluating the clinical predictive performance and credibility of nomograms. ROC curves of OS (A) and BCSS (B) in the LABC group. ROC curves of OS (C) and BCSS (D) in the MBC group. ROC curves of OS (E) and BCSS (F) in the surgery group. BCSS, breast cancer-specific survival; LABC, locally advanced breast cancer; MBC, metastatic breast cancer; OS, overall survival; ROC, receiver operating characteristic.

Figure 4 Calibration curves evaluating the clinical reliability of nomograms. Calibration curves of OS (A) and BCSS (B) in the LABC group. Calibration curves of OS (C) and BCSS (D) in the MBC group. Calibration curves of OS (E) and BCSS (F) in the surgery group. BCSS, breast cancer-specific survival; LABC, locally advanced breast cancer; MBC, metastatic breast cancer; OS, overall survival.

Figure 5 DCA curves of OS (A) and BCSS (B) in the LABC group. DCA curves of OS (C) and BCSS (D) in the MBC group. DCA curves of OS (E) and BCSS (F) in the surgery group. BCSS, breast cancer-specific survival; DCA, decision curve analysis; LABC, locally advanced breast cancer; MBC, metastatic breast cancer; OS, overall survival.

Discussion

Clinically meaningful survival disparities were observed between groups, with MBC patients exhibiting worse prognoses relative to LABC patients (Figure 6). Univariate survival analysis using the Kaplan-Meier method demonstrated that pathology did not influence OS/BCSS in MBC. N-stage lacked prognostic value for BCSS, and bone metastasis did not independently predict OS (Table 5). In a study by Gong et al. on the effect of molecular subtypes on patients with MBC, regional nodes positive had no effect on either OS or BCSS in patients with stage IV breast cancer, which is slightly different from our results (13).

Figure 6 Kaplan-Meier curves were plotted to analyze OS and BCSS. Comparisons of OS (A) and BCSS (B) between patients with LABC and MBC. Comparisons of OS (C) and BCSS (D) between patients in the surgery group and non-surgery group. BCSS, breast cancer-specific survival; LABC, locally advanced breast cancer; MBC, metastatic breast cancer; OS, overall survival.

Breast cancer incidence rates among Caucasian women in Western populations exhibit a unimodal pattern of continuous growth (14). The median age at breast cancer diagnosis among women in the United States is projected to be 62 years (15). Therefore, in our study, U.S. cases from the SEER database were stratified into pre- and post-60 age groups to align with the median age distribution observed in this population. In contrast, breast cancer incidence rates among Chinese women exhibit a bimodal distribution: 45–55 years (perimenopausal period) and 65–75 years (postmenopausal period) (16,17). This unique demographic distribution underscores the need for age-stratified screening and treatment strategies in Chinese populations, which may differ significantly from unimodal patterns observed in Western cohorts.

In the SEER database, tumor grade for cases diagnosed in 2017 or earlier is performed according to North American Association of Central Cancer Registries (NAACCR) criteria as follows: G1-well differentiated, G2-moderately differentiated, G3-poorly differentiated, G4-undifferentiated/anaplastic. Despite G4 tumors representing the highest histological grade, their disproportionately low prevalence (0.2%) in our dataset may reflect bias, leading to attenuated nomogram scores for this variable. Because G4 tumors are biologically indistinguishable from G3 neoplasms in key parameters including proliferative activity, invasiveness, and metastatic potential, current clinical guidelines reclassify G3 and G4 tumors under a unified “high-grade” category (18). The T-stage classification of the 6th edition AJCC staging system is based on tumor size and local invasion depth, as defined below: T1: tumor with maximum diameter ≤2 cm; T2: tumor diameter >2 and ≤5 cm; T3: tumor diameter >5 cm; T4: tumor invades chest wall and/or skin (manifestations include skin ulceration, cellulitis, and satellite nodules). In the surgery cohort, all T1 and T2 patients developed distant metastasis, whereas only 1,247 (9.6%) of T3/T4 patients exhibited metastatic spread. Consequently, T3 patients demonstrated better survival outcomes in the survival analysis and received lower scores on the nomograms. This finding reminds us to focus on not only evaluating local tumor burden but also assessing for concurrent distant metastases in patients with ABC, while emphasizing the importance of systemic health management and integrated surveillance strategies.

While univariate analysis suggested a potential prognostic benefit of radiotherapy, multivariate Cox regression failed to demonstrate statistical significance for OS or BCSS, consistent with the findings of Gong et al. (13) and Lang et al. (19). Historically, a substantial body of research has been dedicated to evaluating the role of surgical interventions in MBC, while fewer studies have dealt with the efficacy of radiotherapy alone (20). Multiple retrospective analyses have reported inconsistent findings regarding this topic. Rhu et al. reported no significant survival benefit associated with radiotherapy delivered to the primary tumor and/or metastatic lesions in patients with MBC (21). Conversely, Le Scodan et al. (22) and Bourgier et al. (23) reported a statistically significant association between radiotherapy monotherapy and improved long-term OS. Additionally, Akay et al. (24) demonstrated that radiotherapy was associated with improved OS and distant progression-free survival (DPFS) in patients with inflammatory breast cancer. The multivariable model included clinically relevant covariates, including age, race, tumor grade, T-stage, surgery, chemotherapy exposure, ER/PR/HER2 status, and metastasis site. Notably, several of these variables are potential confounding factors in radiotherapy analysis, as patients selected for radiotherapy tend to exhibit younger age, advanced tumor stage, and receipt of multimodal therapy, which may confound the observed association through indication bias. Furthermore, comorbidities (e.g., cardiovascular disease, diabetes mellitus) and treatment adherence (e.g., whether patients complete the full course of radiotherapy or chemotherapy) both influence survival. However, these variables—which are correlated with both radiotherapy receipt and survival outcomes—are not captured in the SEER database, resulting in the overestimation of radiotherapy’s apparent survival relevance in univariate analyses of the MBC cohort. These results underscore to clinicians the importance of tailored treatment strategies for metastatic disease: oligometastatic patients may derive significant benefit from radiotherapy to achieve local tumor control, thereby improving both survival duration and quality of life; whereas patients with extensive metastases should primarily rely on systemic therapeutic modalities (e.g., chemotherapy, targeted therapy, immunotherapy) to manage disease burden.

The traditional view is that patients with MBC (stage IV) should be treated with systemic therapy, and surgery is only used to relieve local symptoms such as tumor bleeding, pain, or ulcers, and to improve the quality of life (25). At the 19^th St. Gallen International Breast Cancer Conference, in a vote on whether a patient with limited metastatic disease and highly effective treatment options and/or good initial response to therapy one should strongly consider definitive locoregional treatment, 87.1% of the panel chose to agree. Several prospective randomized controlled trials (10-12) demonstrate that cytoreductive surgery does not improve OS in MBC patients, aligning with systemic therapy as the primary treatment strategy. Notwithstanding, emerging evidence indicates that surgical treatment is associated with improved long-term survival outcomes compared with systemic monotherapy alone (1,8,9,26). Our analysis demonstrated survival benefit associated with surgical intervention across all patient groups evaluated. Within the MBC cohort of this study, patients who underwent partial mastectomy or mastectomy achieved prolonged survival compared with nonsurgical counterparts (mean OS: 71 vs. 64 vs. 41 months, P<0.001; mean BCSS: 77 vs. 69 vs. 46 months, P<0.001, respectively) (Table 5). Similarly, the benefit of partial mastectomy was best in the LABC group, followed by mastectomy and non-surgery (mean OS: 109 vs. 101 vs. 67 months, P<0.001; mean BCSS: 121 vs. 110 vs. 81 months, P<0.001, respectively) (Table 3). This observation may be attributed to the higher utilization of adjuvant radiotherapy and chemotherapy in BCS patients, which is linked to improved long-term outcomes compared to mastectomy cohorts. The landmark NSABP B-06 trial demonstrated no statistically significant differences in OS, disease-free survival, or long-term disease-free survival between women treated with total mastectomy versus breast-conserving therapy (BCT) consisting of lumpectomy with or without adjuvant radiotherapy after 20 years of follow-up (27). The EORTC 10801 trial compared BCT with modified radical mastectomy (MRM); long-term follow-up showed similar survival between two groups (28). These findings underscore to clinicians that treatment selection for breast cancer patients should prioritize multimodal systemic therapy over extensive surgical resection, in accordance with National Comprehensive Cancer Network (NCCN) guidelines advocating integrated oncologic care. In addition, our study indicated that patients in the surgery group exhibited significantly higher 3-, 5-, and 10-year OS and BCSS compared with those in the non-surgery group (Figure 6).

In the MBC group, bone metastases were documented in 3,918 patients (63.6%), with secondary metastatic sites identified in descending order as the lung (1,883, 30.6%), liver (1,499, 24.3%), and brain (403, 6.5%) (Table 1). Previous studies have shown that bone is the most common distant metastatic organ in breast cancer patients (29,30). Nevertheless, beyond bone metastases, involvement of extraosseous metastatic sites independently correlates with worse prognosis, with significant differences observed across specific organ systems (Table 6). Brain metastases are associated with worse clinical outcomes due to the limited penetration of first-line systemic agents across the blood-brain barrier (31). Our findings are concordant with Gong et al.’s (13) and Wang et al.’s (32) observations, demonstrating that patients with bone metastases exhibited the most favorable prognosis (mean OS =51 months, 95% CI: 49.029–51.996), whereas those with brain metastases demonstrated the poorest survival outcomes (mean OS =24 months, 95% CI: 20.085–26.935); comparable survival outcomes were noted between liver (mean OS =40 months, 95% CI: 37.454–42.070) and lung (mean OS =41 months, 95% CI: 38.544–42.628) metastases. However, Gerratana et al. reported that breast cancer patients with lung as the primary site of distant metastasis demonstrated the longest median OS at 58.5 months, which was prolonged compared with those presenting with bone metastasis (44.4 months) (33). Factors contributing to this discrepancy include the fact that this study enrolled more triple-negative breast cancer (TNBC) patients, who have a predilection for lung metastases and gain survival benefit from chemotherapy. However, multivariate analysis in the surgery group revealed a paradoxical finding: the hazard ratio (HR) for lung metastasis was numerically lower than that of bone metastasis in both OS (HR: 1.263, 95% CI: 1.131–1.410 vs. HR: 1.281, 95% CI: 1.159–1.415) and BCSS (HR: 1.328, 95% CI: 1.181–1.493 vs. HR: 1.393, 95% CI: 1.254–1.549) (Table S3), and lung metastasis demonstrated lower scores in the nomograms (Figure 2E,2F). Breast cancer lung metastasis is frequently characterized by clinical manifestations including pain, cough, hemoptysis, pleural effusion, and pulmonary dysfunction (34). These symptoms are often brought to the attention of both patients and clinicians. Breast cancer bone metastases typically manifest as a constellation of skeletal-related events, including pain, pathologic fractures, hypercalcemia, and spinal cord compression (35). Bone pain is the most common complication of metastatic bone disease, reported in 81.4% of patients with metastatic cancer (36). In hormone receptor-positive breast cancer patients with high propensity for bone metastases, aromatase inhibitors (AIs) are frequently prescribed as adjuvant therapy. Paradoxically, AI-induced skeletal adverse effects may obscure these symptoms above, potentially delaying the detection of bone metastases. The diagnosis of bone metastasis is based mainly on imaging techniques, including X-ray imaging, skeletal scintigraphy (SS), CT, MRI, and FDG-PET/CT (35). Conventional X-ray exhibits limited sensitivity for detecting osseous metastases, while SS can only identify osteoblastic lesions. CT requires metastatic deposits ≥1 cm in size for visualization, whereas MRI and PET-CT are more accurate, making early diagnosis of breast cancer bone metastases is more complex (37).

Our results confirmed that HR⁺ and HER2⁺ patients demonstrated superior long-term survival outcomes compared with other molecular subtypes. Breast cancer molecular subtypes exhibit distinct site-specific metastatic tropisms, with variable predilection for visceral, osseous, or central nervous system involvement. Several studies have analyzed the influence of molecular subtypes on MBC in detail. Gong et al.’s results indicate that the prognosis among the different molecular subtypes of stage IV breast cancer patients is highly variable. Specifically, median OS and BCSS were approximately 3.5 times higher in HER2⁺/HR⁺ patients than in HER2⁻/HR⁻ patients, while survival was similar in HER2⁻/HR⁺ and HER2⁺/HR⁻ patients (26). HR⁺ breast cancer patients exhibit a statistically significant increased risk of developing bone metastases, with this association consistently observed across multiple researches (32,38,39). HER2 status is strongly associated with brain metastasis (40). However, considerable debate persists regarding visceral metastasis. Liver metastasis was most frequent in HER2⁺/HR⁻ patients, whereas lung metastasis was most common in HER2⁻/HR⁻ patients (26,41). However, Park et al. concluded that liver metastasis is not associated with molecular subtypes (41).

However, there are certain limitations in this study. The study was based exclusively on data from the SEER database (a single U.S.-based cancer registry), and we did not perform external validation due to the limited sample size available from a single clinical center. While internal validation confirmed the model’s reliability within the SEER cohort, external validation is still needed to verify whether the nomograms perform consistently in diverse clinical settings, ethnic groups, or healthcare systems. To mitigate this limitation, we plan to conduct external validation in subsequent research via two complementary strategies: collaborating with domestic and international medical centers to collect multicenter retrospective/prospective data from ABC patients, covering diverse ethnic backgrounds and clinical treatment paradigms; utilizing publicly accessible independent datasets (e.g., TCGA-BRCA, METABRIC) to assess the model’s discriminative performance and calibration accuracy. We expect that this external validation will refine the proposed nomograms and enhance their reliability to support clinical decision-making across varied clinical contexts.

Conclusions

In conclusion, this study systematically characterized the clinicopathological features of patients with ABC, evaluated the prognostic impact of multiple factors (including clinicopathological characteristics and treatment modalities), and developed prognostic nomograms to guide precision treatment decisions for clinicians.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2588/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2588/prf

Funding: The research was supported by The First Affiliated Hospital of Harbin Medical University Fund for Distinguished Young Medical Scholars (No. 2021J17) and Beijing Medical Award Foundation (No. YXJL-2021-0302-0287). The funders did not participate in study design, data collection, data analysis, or manuscript preparation, and no conflicts of interest exist.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2588/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, Giaquinto AN, Jemal A. Cancer statistics, 2024. CA Cancer J Clin 2024;74:12-49. [Crossref] [PubMed]
2. Benitez Fuentes JD, Morgan E, de Luna Aguilar A, et al. Global Stage Distribution of Breast Cancer at Diagnosis: A Systematic Review and Meta-Analysis. JAMA Oncol 2024;10:71-8. [Crossref] [PubMed]
3. Cardoso F, Paluch-Shimon S, Schumacher-Wulf E, et al. 6th and 7th International consensus guidelines for the management of advanced breast cancer (ABC guidelines 6 and 7). Breast 2024;76:103756. [Crossref] [PubMed]
4. Breast Cancer Expert Committee of National Cancer Quality Control Center. Cancer Drug Clinical Research Committee of China Anti-Cancer Association. Zhonghua Zhong Liu Za Zhi 2024;46:1079-106.
5. Liedtke C, Kolberg HC. Systemic Therapy of Advanced/Metastatic Breast Cancer - Current Evidence and Future Concepts. Breast Care (Basel) 2016;11:275-81. [Crossref] [PubMed]
6. Carson E, Dear R. Advanced breast cancer: An update to systemic therapy. Aust J Gen Pract 2019;48:278-83. [Crossref] [PubMed]
7. Retamales J, Daneri-Navarro A, Artagaveytia N, et al. Implementing Standard Diagnosis and Treatment for Locally Advanced Breast Cancer Through Global Research in Latin America: Results From a Multicountry Pragmatic Trial. JCO Glob Oncol 2024;10:e2300216. [Crossref] [PubMed]
8. Soran A, Ozmen V, Ozbas S, et al. Primary Surgery with Systemic Therapy in Patients with de Novo Stage IV Breast Cancer: 10-year Follow-up; Protocol MF07-01 Randomized Clinical Trial. J Am Coll Surg 2021;233:742-751.e5. [Crossref] [PubMed]
9. Si Y, Yuan P, Hu N, et al. Primary Tumor Surgery for Patients with De Novo Stage IV Breast Cancer can Decrease Local Symptoms and Improve Quality of Life. Ann Surg Oncol 2020;27:1025-33. [Crossref] [PubMed]
10. King TA, Lyman JP, Gonen M, et al. Prognostic Impact of 21-Gene Recurrence Score in Patients With Stage IV Breast Cancer: TBCRC 013 J Clin Oncol 2016;34:2359-65. [Crossref] [PubMed]
11. Fitzal F, Bjelic-Radisic V, Knauer M, et al. Impact of Breast Surgery in Primary Metastasized Breast Cancer: Outcomes of the Prospective Randomized Phase III ABCSG-28 POSYTIVE Trial. Ann Surg 2019;269:1163-9. [Crossref] [PubMed]
12. Badwe R, Hawaldar R, Nair N, et al. Locoregional treatment versus no treatment of the primary tumour in metastatic breast cancer: an open-label randomised controlled trial. Lancet Oncol 2015;16:1380-8. [Crossref] [PubMed]
13. Gong Y, Liu YR, Ji P, et al. Impact of molecular subtypes on metastatic breast cancer patients: a SEER population-based study. Sci Rep 2017;7:45411. [Crossref] [PubMed]
14. Sung H, Rosenberg PS, Chen WQ, et al. Female breast cancer incidence among Asian and Western populations: more similar than expected. J Natl Cancer Inst 2015;107:djv107. [Crossref] [PubMed]
15. Siegel RL, Miller KD, Fuchs HE, et al. Cancer statistics, 2022. CA Cancer J Clin 2022;72:7-33. [Crossref] [PubMed]
16. Sun K, Zhang B, Lei S, et al. Incidence, mortality, and disability-adjusted life years of female breast cancer in China, 2022. Chin Med J (Engl) 2024;137:2429-36. [Crossref] [PubMed]
17. Fan L, Strasser-Weippl K, Li JJ, et al. Breast cancer in China. Lancet Oncol 2014;15:e279-89. [Crossref] [PubMed]
18. Byrd DR, Brookland RK, Washington MK, et al. AJCC cancer staging manual. New York: Springer; 2017.
19. Lang JE, Tereffe W, Mitchell MP, et al. Primary tumor extirpation in breast cancer patients who present with stage IV disease is associated with improved survival. Ann Surg Oncol 2013;20:1893-9. [Crossref] [PubMed]
20. Janssen S, Rades D. Primary Breast Cancer with Synchronous Metastatic Disease - Indications for Local Radiotherapy to the Breast and Chest Wall. Anticancer Res 2015;35:5807-12. [PubMed]
21. Rhu J, Lee SK, Kil WH, et al. Surgery of primary tumour has survival benefit in metastatic breast cancer with single-organ metastasis, especially bone. ANZ J Surg 2015;85:240-4. [Crossref] [PubMed]
22. Le Scodan R, Stevens D, Brain E, et al. Breast cancer with synchronous metastases: survival impact of exclusive locoregional radiotherapy. J Clin Oncol 2009;27:1375-81. [Crossref] [PubMed]
23. Bourgier C, Khodari W, Vataire AL, et al. Breast radiotherapy as part of loco-regional treatments in stage IV breast cancer patients with oligometastatic disease. Radiother Oncol 2010;96:199-203. [Crossref] [PubMed]
24. Akay CL, Ueno NT, Chisholm GB, et al. Primary tumor resection as a component of multimodality treatment may improve local control and survival in patients with stage IV inflammatory breast cancer. Cancer 2014;120:1319-28. [Crossref] [PubMed]
25. Khan SA, Stewart AK, Morrow M. Does aggressive local therapy improve survival in metastatic breast cancer? Surgery 2002;132:620-6; discussion 626-7. [Crossref] [PubMed]
26. Liu X, Wang C, Feng Y, et al. The prognostic role of surgery and a nomogram to predict the survival of stage IV breast cancer patients. Gland Surg 2022;11:1224-39. [Crossref] [PubMed]
27. Fisher B, Anderson S, Bryant J, et al. Twenty-year follow-up of a randomized trial comparing total mastectomy, lumpectomy, and lumpectomy plus irradiation for the treatment of invasive breast cancer. N Engl J Med 2002;347:1233-41. [Crossref] [PubMed]
28. Litière S, Werutsky G, Fentiman IS, et al. Breast conserving therapy versus mastectomy for stage I-II breast cancer: 20 year follow-up of the EORTC 10801 phase 3 randomised trial. Lancet Oncol 2012;13:412-9. [Crossref] [PubMed]
29. Wu SG, Sun JY, Yang LC, et al. Patterns of distant metastasis in Chinese women according to breast cancer subtypes. Oncotarget 2016;7:47975-84. [Crossref] [PubMed]
30. Kennecke H, Yerushalmi R, Woods R, et al. Metastatic behavior of breast cancer subtypes. J Clin Oncol 2010;28:3271-7. [Crossref] [PubMed]
31. Lin NU, Amiri-Kordestani L, Palmieri D, et al. CNS metastases in breast cancer: old challenge, new frontiers. Clin Cancer Res 2013;19:6404-18. [Crossref] [PubMed]
32. Wang R, Zhu Y, Liu X, et al. The Clinicopathological features and survival outcomes of patients with different metastatic sites in stage IV breast cancer. BMC Cancer 2019;19:1091. [Crossref] [PubMed]
33. 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]
34. Jin L, Han B, Siegel E, et al. Breast cancer lung metastasis: Molecular biology and therapeutic implications. Cancer Biol Ther 2018;19:858-68. [Crossref] [PubMed]
35. Pang L, Gan C, Xu J, et al. Bone Metastasis of Breast Cancer: Molecular Mechanisms and Therapeutic Strategies. Cancers (Basel) 2022;14:5727. [Crossref] [PubMed]
36. Cleeland CS, Portenoy RK, Rue M, et al. Does an oral analgesic protocol improve pain control for patients with cancer? An intergroup study coordinated by the Eastern Cooperative Oncology Group. Ann Oncol 2005;16:972-80. [Crossref] [PubMed]
37. Lecouvet FE, Larbi A, Pasoglou V, et al. MRI for response assessment in metastatic bone disease. Eur Radiol 2013;23:1986-97. [Crossref] [PubMed]
38. Kai M, Kogawa T, Liu DD, et al. Clinical characteristics and outcome of bone-only metastasis in inflammatory and noninflammatory breast cancers. Clin Breast Cancer 2015;15:37-42. [Crossref] [PubMed]
39. Liede A, Jerzak KJ, Hernandez RK, et al. The incidence of bone metastasis after early-stage breast cancer in Canada. Breast Cancer Res Treat 2016;156:587-95. [Crossref] [PubMed]
40. Palmieri D, Bronder JL, Herring JM, et al. Her-2 overexpression increases the metastatic outgrowth of breast cancer cells in the brain. Cancer Res 2007;67:4190-8. [Crossref] [PubMed]
41. 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]