1 Introduction

Breast cancer is the most commonly diagnosed cancer and the leading cause of cancer-related mortality among women worldwide [1]. The majority of breast cancer patients present with localized disease, primarily confined to the breast and adjacent lymph nodes (stages I–III), with current treatment protocols offering extended survival for most [2, 3]. However, despite significant therapeutic advancements in recent decades, the 5-year survival rate for patients with metastatic breast cancer remains discouragingly low, at approximately 32% [1]. Understanding the characteristics of this critical subgroup is essential for improving clinical surveillance and developing more effective management strategies.

The most common sites of metastasis in breast cancer include the bone, liver, lung, and brain. Several factors, such as tumor size, lymph node involvement, histological grade, pathological subtype, and key biological markers such as hormone receptor and HER2 status, influence the onset and progression of metastatic disease. Wang et al. [4] identified significant differences in distant metastasis patterns among patients with various breast cancer subtypes, noting that those with brain metastasis had the poorest cancer-specific survival across all subtypes. These findings were corroborated by Xiao et al. [5]. Although molecular subtypes play a crucial role in predicting site-specific metastases, recent studies have highlighted the importance of other factors in influencing metastatic patterns. Wang et al. [6] demonstrated that primary tumor location affects metastatic spread, with upper-outer quadrant tumors associating with bone metastases and central tumors showing a higher propensity for liver metastases. Additionally, sociodemographic factors such as race [7] and marital status [8] have been found to modulate metastatic risk. Notably, patients with lower socioeconomic status tend to have higher rates of metastatic disease at diagnosis and are more likely to present with multiple metastatic sites. Despite these insights, the complex interplay between sociodemographic factors, clinicopathological variables, and distant metastasis patterns remains poorly understood. Furthermore, the pattern of multisite metastases, a critical aspect of advanced breast cancer, has yet to be comprehensively explored. This gap underscores the need for more granular investigations into the multifaceted nature of metastatic breast cancer.

In recent years, breast cancer research has expanded beyond the study of localized disease to address the complexities of metastatic breast cancer. This shift reflects an increasing recognition of the urgent need to understand the mechanisms of metastasis, which remains the leading cause of death in breast cancer patients. Leveraging population-based data from the SEER database, the current study aims (1) to describe the incidence and composite ratio according to major distant metastasis sites, and (2) to investigate the survival and the associations between major clinicopathological variables and different metastasis pattern. By analyzing data from diverse patient demographics and cancer characteristics, we aim to uncover patterns that could inform more personalized and effective treatment approaches.

2 Methods

3 Results

Our comprehensive analysis of 356,789 breast cancer patients provides significant insights into the patterns, risk factors, and prognosis of metastatic breast cancer. This study, one of the largest of its kind, reveals important trends in metastatic site preferences across different molecular subtypes and demographic groups. We uncovered distinct relationships between patient characteristics, tumor biology, and metastatic behavior, which have significant implications for survival outcomes. Our findings highlight the complex interplay between tumor grade, race, age, and metastatic patterns, emphasizing the need for personalized approaches in managing metastatic breast cancer. These results not only confirm some previously known associations but also reveal novel patterns that could inform future research and clinical decision-making in breast cancer care.

The final cohort incorporated 356,789 breast cancer patients (HR+/HER2− 242,541 [68.0%], HR+/HER2+ 35,090 [9.8%], HR−/HER2+ 15,062 [4.2%], TNBC 37,455 [10.5%]), out of which 18,036 (5.06%) were diagnosed with stage IV of the disease. Notably, the molecular subtype distributions align closely with previous scholarly studies, particularly with those concerning US-based patient cohorts [15]. The data set's attributes are bifurcated and elucidated at two analytical levels: the incidence proportion and the composition ratio, examined both across the entire cohort and specifically within the metastatic subset.

3.4 Prognostic Analysis

Table S4 furnishes a detailed portrayal of median survival durations for breast cancer patients displaying specific metastatic patterns, further categorized by sociodemographic and clinicopathological variables. Among these, patients with bone metastasis reported the longest median survival duration (27 months), succeeded by lung (18 months), liver (15 months), and brain metastasis (9 months). The Kaplan–Meier evaluations were bifurcated into two distinct paradigms: (1) construction of survival curves for patients, categorized within specific sociodemographic and clinicopathological variables, and delineated by metastatic patterns; (2) depiction of survival curves for patients, characterized by certain metastatic patterns, and further stratified by sociodemographic and clinicopathological criteria. Both Figures 1 and S2 are emblematic of the aforementioned paradigms. Within the subset manifesting any metastasis, a direct correlation was observed: as the count of metastatic sites escalated, the median survival duration plummeted. Typically, increased age corresponded to reduced survival durations, as substantiated by Table S4. Supplementary Kaplan–Meier visualizations remain undisclosed. The seminal observations are epitomized in Table 1D.

4 Discussion

Prior literature on this topic can generally be categorized into two main approaches: the first focuses on analyzing metastasis patterns and prognosis by isolating individual variables such as subtype or gender. The second emphasizes the role of molecular subtypes in influencing metastasis distribution and prognosis, often concentrating on a specific metastatic site in breast cancer. Using the SEER dataset from 2010 to 2013, Wang et al. [4] found that bone metastasis was the most prevalent, whereas brain metastasis occurred less frequently. Their study highlighted that the HR+/HER2− subtype was most strongly associated with metastasis, whereas the HR−/HER2+ subtype exhibited the lowest metastasis rates. However, their analysis was limited to the molecular subtypes. Similarly, Xiao et al. [5] explored metastatic patterns and prognoses through the lens of subtypes, identifying a strong correlation between HR+/HER2+ and HR−/HER2+ subtypes and an increased risk of metastasis to the liver, brain, and lungs, whereas the HR−/HER2− subtype was inversely associated with bone metastasis. Furthermore, Xie et al. [16] examined metastatic patterns in male breast cancer (MBC) and compared them with female breast cancer (FBC), revealing distinct clinicopathological features and metastatic patterns in metastatic MBC. However, prognostic outcomes between MBC and FBC patients were similar, despite a higher incidence of central breast tumors and older age (≥ 60 years) in MBC patients. Martin et al. [17] focused on the incidence and prognosis of brain metastases, noting higher incidence rates among patients with HR−/HER2+ and HR−/HER2− subtypes. Interestingly, patients with HR+/HER2+ breast cancer had better survival rates, whereas those with HR−/HER2− had the poorest prognosis. A common limitation across these studies is their narrow focus on a single metastatic site or a single variable, which may overlook essential insights that a more comprehensive, comparative analysis could reveal. Moreover, none of these studies accounted for the role of metastasis relapse.

In contrast to most existing literature, our study takes a broader approach by examining all patterns of distant metastasis in breast cancer patients at the time of diagnosis, utilizing a multifaceted analytical perspective supported by a large sample size. Xiao et al.'s research emphasizes that the impact of age on metastatic patterns is not uniform across different sites [5]. In our study, we observed that the impact of age on metastasis varied significantly depending on the metastatic site. Younger patients (< 40 years) exhibited a higher incidence of bone metastasis, which is consistent with previous studies suggesting that younger women often have higher bone mineral density and a more active bone microenvironment, providing a favorable site for cancer cell growth. Additionally, younger patients may have higher estrogen levels, which can promote bone metastasis in hormone receptor-positive tumors. In contrast, older patients (> 70 years) showed a higher propensity for lung metastasis. This could be due to age-related declines in immune function and the increased vulnerability of the lungs to metastatic disease. Furthermore, as the immune system becomes less efficient with aging, the ability to suppress micrometastases may be reduced, allowing tumors to establish in distant organs such as the lungs and liver. Interestingly, the distribution of metastasis to the brain did not show a significant age-related difference, suggesting that brain metastases may be more closely associated with the tumor's biological characteristics (e.g., HER2+ or triple-negative) rather than age. The data indicate that grade III tumors, which are more common in younger patients, have a higher tendency to metastasize to the brain, whereas liver and lung metastases are often more prevalent in older patients, reflecting the increased systemic spread and immune system inefficiency in this population.

Notably, younger breast cancer patients seemed predisposed to a heightened risk of bone metastases, although the variance failed to achieve statistical significance. As age advanced, the proclivity for liver metastases diminished, whereas the susceptibility to lung metastases ascended [5]. Pertaining to brain metastases, individuals within the age bracket of 40–65 years manifested a more pronounced risk in comparison to other age cohorts [5]. This alignment of findings mirrors our own: the influence of age exhibited fluctuations depending on the metastasis site. Elevated age cohorts demonstrated an augmented predisposition for lung metastasis, but a reduced risk for bone and liver metastasis. Age-related associations with brain metastasis did not achieve statistical significance. Xie et al. discerned that, in juxtaposition with non-metastatic MBC patients, those with metastatic MBC portrayed distinct clinicopathological characteristics and deviated from their metastatic FBC counterparts. Nonetheless, prognostic outcomes bore no significant distinctions between metastatic MBC and FBC patients. In our exploration, non-metastatic breast cancer female patients exhibited a more extended survival span compared to their male counterparts (HR 0.8; 95% CI 0.73–0.87; p < 0.001). Concerning metastatic patients, our analysis discerned that the subtype distribution was roughly analogous between genders. However, male patients presented a diminished composition ratio of TNBC (1.8%) in contrast to female patients (10.6%), a finding in concordance with previously cited data [18].

Interestingly, our study revealed that single (unmarried) patients exhibited an elevated incidence of metastasis. This observation aligns with prior literature indicating that unmarried individuals tend to experience poorer survival outcomes in breast cancer [8]. A potential rationale for this phenomenon might be the heightened susceptibility of unmarried patients to psychological distress and engagement in detrimental habits, stemming from the absence of financial stability and emotional support often provided by spouses [19, 20]. Remarkably, after neutralizing the stage effect in multivariate analyses, the heightened metastasis risk resurfaced. However, when mitigating the effect of subtype, the outcome remained largely unchanged. This indicates that delayed diagnosis could be a pivotal determinant of an augmented metastasis risk among black patients. Furthermore, within the entire sample, the incidence of bone metastasis for bilateral breast cancer surpassed that of unilateral cases (57.17% bilateral vs. 3.56% left). However, among those with metastasis, the lung metastasis incidence for bilateral cancer was relatively diminished in comparison to unilateral cases (29.27% bilateral vs. 35.53% left). Distinctively, patients with bilateral breast cancer manifested inferior survival outcomes compared to their counterparts. Such an observation might be elucidated by the fact that (1) bilateral breast cancer patients tend to receive a diagnosis at more advanced stages than those with unilateral manifestations; and (2) a significant proportion of apparent bilateral primary tumors might actually represent metastatic growths from the opposing side. The reduced metastasis incidence in patients undergoing surgery, radiotherapy, or chemotherapy is presumably inversely correlated, as those manifesting overt metastasis may forgo locoregional treatments. Pertaining to bone metastasis, black patients, in contrast to their white counterparts, consistently demonstrated a significantly diminished OR for manifesting bone metastasis at diagnosis. There was an absence of disparity between white and black patients concerning brain and lung metastasis, a finding incongruent with the conclusions drawn by Martin et al. [17]. This deviation necessitates further investigations for elucidation. Our data indicated a predisposition for patients undergoing chemotherapy to develop liver metastases, whereas those receiving radiotherapy exhibited a proclivity for bone metastasis. A plausible explanation for such an occurrence might be the common clinical practice of prescribing radiotherapy for bone metastasis and chemotherapy for liver metastasis.

We observed that uninsured patients exhibited a heightened proportion of metastatic incidence compared to those insured. This trend may be attributed to the uninsured cohort's probable association with an absence of preventative measures, limited screening, restricted access to healthcare, and diagnostic delays, culminating in more advanced disease presentations at diagnosis [21]. There is a dearth of studies examining metastatic incidence amongst patients with bilateral breast cancer. Our findings underscored an elevated metastatic incidence amongst this demographic. A singular precedent study highlighted that patients with bilateral breast cancer bore a 1.25-fold augmented risk of distant metastasis relative to unilateral cases [22]. Given the infrequency of bilateral breast cancer, our metastatic incidence results necessitate judicious interpretation.

For all metastatic patterns, we identified that patients with triple-negative breast cancer (TNBC) manifested the briefest survival span, whereas those with HR+/HER2+ exhibited the most favorable prognosis. Notably, our study pioneers the observation that metastatic site exhibited negligible impact on the survival of TNBC patients but displayed pronounced significance in HR+/HER2−. There was a discernible decrement in survival with each successive age decade, which can be partially attributed to the fact that older individuals might be predisposed to noncancerous mortality causes. Conversely, younger patients might be afforded enhanced access to locoregional therapeutic modalities, encompassing surgery, radiotherapy, and chemotherapy, thereby amplifying their survival prospects [23, 24]. Interestingly, our analysis revealed that patients with HR+HER2+ demonstrated superior survival relative to the HR+HER2− subtype, an outcome potentially influenced by the availability of more precision-targeted therapeutic regimens. In addition to this, recent studies have provided further insights into the molecular mechanisms driving metastatic spread in specific organs. Ganesan et al. [25] demonstrated that the EGFR-mediated PI3K/Akt/mTOR signaling pathway plays a crucial role in promoting lung metastasis in TNBC, highlighting the potential therapeutic targets for lung metastasis, a site more common in older patients in our study. Similarly, Zhang et al. [26] identified as a key regulator of liver metastasis, underscoring the importance of circRNAs in regulating liver metastasis in breast cancer. This is particularly relevant given the increased incidence of liver metastasis in older patients, as observed in our study. Furthermore, Xie et al. [27] utilized single-cell RNA sequencing to map the brain metastasis ecosystem, pinpointing as a promising therapeutic target. Brain metastasis, which was not significantly age-dependent in our study, might benefit from therapies targeting these molecular markers.

This investigation is not devoid of limitations. (1) Screening modalities: Given that breast cancer screening guidelines don't invariably advocate for the employment of the most sensitive screening modalities, it's plausible that our estimates undershoot the actual metastatic rates. (2) Data completeness: The SEER database does not furnish information on residence type, educational attainment, median household income, comorbidities, performance status, smoking habits, psychological health, or pivotal genomic data. These missing factors could influence the generalizability of our findings, especially when considering the diversity of factors that affect cancer outcomes. (3) Demographic representation: The SEER database includes data from 18 cancer registries in the United States, covering approximately 28% of the US population. Although this provides valuable insights, it may not fully represent the cancer characteristics of different demographic groups, especially those from non-US populations or under-represented groups. Additionally, the database relies on self-reported demographic information, which could introduce reporting biases, particularly with regard to race and marital status. (4) Metastatic details and treatment information: The SEER database does not delineate the precise metastatic locations, metastatic count, or exhaustive treatment details. As a result, we were unable to fully explore the complexity of multisite metastasis or the impact of specific therapies on metastatic spread. (5) De novo metastasis data: This study exclusively hinged on de novo metastasis data, eschewing data on relapsed metastatic cancer, and relying solely on overall survival metrics. The exclusion of relapsed cancer data may limit the applicability of our findings to patients who experience later-stage metastasis following initial treatment. (6) Retrospective design and data limitations: As a retrospective cohort study, this investigation has inherent limitations, including the inability to establish causal relationships and the potential for bias in data collection. Although the SEER database provides valuable population-level data, its retrospective nature means that the data were not collected with a specific research hypothesis in mind, which may limit the ability to control for all potential confounding factors. Furthermore, the lack of detailed clinical data (such as treatment regimens, performance status, or comorbidities) means that the influence of certain confounders on metastatic progression and survival outcomes may not be fully captured. These limitations could impact the generalizability of our findings, particularly in more specific patient subgroups, such as those with complex clinical conditions or those receiving specific treatments not recorded in the SEER dataset.

The variation in metastatic patterns among different molecular subtypes, such as HR+/HER2+ and HR−/HER2−, contributes critical insights into treatment decisions. Patients with HR+/HER2+ breast cancer, for example, who are more likely to develop bone and lung metastases, may benefit from combination therapies targeting both HER2 receptors and hormonal pathways. In contrast, those with HR−/HER2− subtypes, who are more prone to lung and liver metastases, may require different systemic therapies. These findings suggest the need for a more patient-centered approach to treatment, where the clinical team, including oncologists, radiologists, and surgeons, collaborate to develop individualized treatment plans that integrate both systemic and localized therapies based on the patient's specific metastatic profile. These personalized strategies are not only likely to improve clinical outcomes but could also help make cancer care more cost-effective. By focusing on high-risk groups, healthcare systems can allocate resources more efficiently, avoiding overuse of expensive treatments in lower risk populations and improving the management of patients who need more intensive interventions.

Based on the findings of this study, several important directions for future research are warranted. (1) Multicenter, prospective studies: Our study, being retrospective in nature, lays the foundation for further prospective research. A multicenter, longitudinal study could help confirm and expand our findings, particularly in populations outside the United States and those underrepresented in SEER. This would also allow for a more detailed exploration of how different geographic regions or healthcare systems influence metastatic patterns and survival outcomes. (2) Molecular biomarkers and genomic studies: Given the role of biological factors in metastatic spread, future studies should focus on identifying specific molecular biomarkers that contribute to metastasis. Exploring the genetic and epigenetic landscape of breast cancer could provide critical insights into the molecular mechanisms underlying metastatic dissemination and help identify potential targets for therapeutic intervention. (3) Relapsed metastatic disease: Our study focused exclusively on de novo metastatic cases. Future research should include relapsed metastatic disease, as this patient group may exhibit distinct metastatic behaviors and survival outcomes. Investigating the differences between primary and recurrent metastasis could inform tailored treatment strategies. (4) Impact of specific therapies on metastasis: Further research should explore how specific treatments (e.g., chemotherapy, radiotherapy, and targeted therapies) influence metastatic patterns. This could involve examining the effect of treatments on specific metastatic sites (such as bone or brain) and evaluating the long-term outcomes of these therapies in preventing or controlling metastasis. (5) Age-related differences in metastatic patterns: Given the age-related differences observed in metastatic patterns, future studies could investigate the underlying mechanisms driving these differences, including immune system alterations and hormonal influences. Specifically, studies comparing metastatic behavior in younger versus older patients could provide valuable insights into how age affects metastatic dissemination and response to treatment.

5 Conclusions

This study provides meaningful insights into the heterogeneous metastatic patterns of breast cancer, emphasizing the influence of age and molecular subtype on the distribution and prognosis of metastatic disease. Our findings highlight the necessity of adopting a more personalized management to both screening and treatment. Stratified screening strategies—such as prioritizing bone surveillance in younger patients and focusing on lung and liver assessments in older patients—may facilitate earlier detection and more effective interventions. Moreover, treatment plans should be tailored according to the specific metastatic profiles and tumor biology of each patient. Personalized therapeutic strategies, including the use of bone-targeted agents or localized therapies for brain metastases, can significantly improve clinical outcomes. These findings support the implementation of age- and subtype-specific protocols in managing metastatic breast cancer to optimize patient care and healthcare resource utilization.

Author Contributions

Xiangyi Kong: conceptualization (lead), investigation (lead), writing – original draft (equal). Qiang Liu: data curation (equal), formal analysis (lead), methodology (lead). Zheng Qu: investigation (equal), writing – original draft (equal), writing – review and editing (equal). Xiangyu Wang: investigation (equal), project administration (equal), software (equal). Wenxiang Zhang: methodology (equal), resources (equal), visualization (equal). Yulu Liu: conceptualization (equal), project administration (equal), software (equal). Robert Coleman: conceptualization (equal), supervision (equal), visualization (equal). Chunqing Lin: supervision (equal), validation (equal). Jing Wang: funding acquisition (equal), supervision (lead), validation (lead).

Ethics Statement

The analysis followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. Ethical review and approval were waived for this study because the SEER Program provides secondary and open databases.

In accordance with the Declaration of Helsinki, the research was registered in Prospero (https://www.crd.york.ac.uk/prospero/).

Consent

Informed consent was waived due to the use of de-identified data.

Conflicts of Interest

The authors declare no conflicts of interest.