Breast cancer (BC) is the cancer with the highest prevalence worldwide (1) and the leading cause of cancer-related deaths among females (2). BC has a tendency for distant metastasis (3), and the majority of BC-related deaths are due to metastasis (4). BC exhibits heterogeneity in metastasis and prognosis (1). Even though patients with distant metastases are all defined as MBC (5), different sites of metastasis have variable impacts on clinical outcomes (6), and the prognosis varies greatly. The metastatic sites should be taken into consideration when assessing prognosis and making therapeutic strategies for patients with MBC (5).

The lung and pleura are among the most common metastatic sites of BC (7). Cummings MC et al. performed an autopsy examination of women who died of BC and found that the most common organs involved were lung/pleura, followed by bone, liver and non-axillary lymph nodes (8). The lung is generally accepted as one of the primary target visceral organs of BC metastasis (1), and is anatomically related to pleura. Lung metastasis (LM) is the most common accompanied organ metastasis site of pleural metastasis (PM) in BC. Some studies on MBC did not distinguish between LM and PM (8–11). Thus when interpreting these data, it must be noted that LM are referred to as either including or excluding PM (12). The site of first distant metastasis correlates with the survival of patients with BC (13, 14). And lung metastasis (LM) is of particular attention because of its high morbidity and association with high mortality of patients (15). PM usually manifested as pleural nodulations or pleural thickening (16), with or without malignant pleural effusion (MPE) (17). However, PM often goes unnoticed until the appearance of MPE. There is little evidence regarding the prognosis of patients with MBC when pleura is the first recurrence site. In the present study, we wanted to explore the prognostic differences between pleural and lung when serving as the first site of distant metastasis after radical surgery for primary BC, which could help to supplement the vacancy of current data.

MPE is a common manifestation of PM (18) and a frequent complication during the course of MBC (19). Approximately 11% of the patients with BC eventually present with symptomatic pleural effusions; at autopsy, 36%–65% of patients retrospectively suffered from this condition (20, 21). Although MPE is rarely the initial manifestation of cancer (20), it carries a significant symptom burden (22) and is considered to be associated with a dismal prognosis (23). However, not all PM was accompanied by MPE (24), especially when the initial diagnosis of MBC. Therefore, we hypothesized that among BC patients with PM, the presence or absence of MPE would lead to different prognoses. The diversity in prognosis of BC patients is caused by the combined effect of multiple pathological factors. An understanding of prognostic factors is imperative for individualization of prognosis in patients with BC. However, the prognostic factors in BC patients with PM are unclear, particularly when no other distant metastases exist. In addition, the prognosis plays a central role for patients with BC and oncologists to choose optimally treatment in this era of individualized therapy (25). Therefore, an accurate prediction model is needed for this subset of patients.

Over the past few years, nomograms have been widely recognized as a predictive method for several diseases, including BC (26). Nomograms can generate an individual probability of a clinical event by integrating diverse determinant variables and meet the requirement for biologically and clinically integrated models (27). Evidence-based guidelines suggest using conservative treatments in patients with limited life expectancy, whereas they suggest offering more aggressive treatment modalities for patients with better prognosis. Real-world data can inform the outcome comparisons (28). Our study aimed to investigate whether patients with BC in whom PM was the primary event of recurrence exhibit a prognosis different than those with LM. Furthermore, we explored the prognostic factors and created a nomogram model, which can aid physicians in better evaluating the patient’s prognosis and selecting patients for different treatment tactics.

Tumor metastasis contributes to high cancer mortality (30), BC has variable aggressiveness and a high propensity to develop distant metastases (31). Extensive studies have proven that BC exhibits metastatic heterogeneity with distinct metastatic precedence to various organs, leading to differences in responses to therapy and prognoses (1). Recent studies have revealed that BC subtypes differ not only in primary tumor characteristics but also in their metastatic behavior (32). In our study, although there was no significant difference in the molecular subtype of primary cancer among the three groups, PM, LM, and PM combined with LM mainly originated from luminal B (39.7%), triple negative (30.9%), and luminal B (34.9%) types, respectively.

The first site of distant metastasis is associated with the prognosis of BC patients (6). Although pleura is a common metastatic site of BC (7), PM has rarely been reported as the first metastatic site in patients with BC. The proportion of such patients may be underestimated because of the time lag in follow-up examinations or the lack of accurate and effective means of examination. In our research, 29.3% of patients did not visit the hospital until presenting with symptoms related to PM. The prognosis of BC patients with single-site metastasis was significantly better than that of patients with multiple metastatic sites (33). In addition, the presence of visceral metastases has a significant negative prognostic impact on patients (28). Schröder J et al. revealed that patients with bone-only metastasis showed better survival than visceral with or without bone metastases (34). Our results indicated that the prognosis of patients with PM not complicated by other distant sites is indeed better than that of patients combined with LM or whose lung serve as the single distant metastatic site. Despite no significant advantage was observed in M-OS for patients with only LM compared with patients with combined PM, the survival rates at 3-, 5- years were all superior. Despite improvements in treatment, MBC has a poor prognosis and an overall 5-year survival rate of only 27% for patients in the United States (35). However, LM has a relatively good prognosis in visceral metastasis as the first distant metastasis of BC (6, 13). Redig AJ et al. tested the relationship between site of metastasis and outcome, and the best prognosis was observed among patients with lung as first anatomic site of distant metastasis, followed by those with first metastatic involvement of bone, liver and central nervous system (6). Combined the existing data, it might be inferred that PM has a better prognosis than visceral metastasis. However, further validation with clinical data is required and the underlying mechanism should be elucidated.

PM most commonly originates from metastatic lung carcinomas and breast carcinomas (36); the mechanisms include hematogenous spread, direct invasion from a neighboring cancer, and retrograde lymphatic spread from the mediastinum (37). Breast carcinoma is the most common metastatic malignancy identified in pleural effusion specimens from women (38). PM is often accompanied by MPE (39), but not all tumors metastasizing to the pleura cause MPE (40). In our results, the incidence of MPE in patients with PM (81.0% in group A and 76.7% in group C) was high. On the one hand, MPE is consider an unfavorable complication that restricts life quality (41) and related to poor prognosis (42). Consistently, our study reported that in BC, MPE was an independent risk factor for patients with PM. On the other hand, Poe RH et al. reported that the median survival of BC patients in whom MPE was the initial and only recurrent site was 48 months, compared with 12 months for patients associated with other metastatic diseases (43). The MPE is more commonly unilateral and ipsilateral to the primary BC (18), Poe RH et al. suggested that this indicated that MPE was a regional rather than systemic disease, probably accounting for the better outlook in patients with effusion alone (43). Similarly, our data showed that the majority of initial PM was located ipsilateral and had a better prognosis. PM located contralateral/bilateral to the primary BC is a factor that worsens the prognosis. Differently, the patients with MPE without other distant metastases at the initial diagnosis of MBC exhibited a better prognosis compared with LM patients with or without PM, but without significant difference ( Supplemental Figure 2 ). Whereas the M-OS of PM patients without MPE was substantially longer (median M-OS:89.3 months in group A; 64.0 months in group C). Thomas et al. (44) speculated that in BC, the laterality of PM is because of lymphatic dissemination. Similarly, Agalioti T et al. (39) stated that BC may invade the pleura because of local proximity rather than through the bloodstream. This may be one of the reasons for its better prognosis than other distant metastases. Moreover, pleura is of itself innocuous and once thought to be biologically inert (45). Oncogene signals and/or transcription factor activation in tumor cells determine paracrine gene expression. The balance between vasoactive mediators and possible protective molecules in the pleural space dictates the occurrence of vasoactive signaling with subsequent MPE development. In turn, this signal cocktail exert a multitude of effects on tumor cells (46). To some extent, tumor colonization of the pleura but not causing MPE may be a manifestation of its poor malignant biological behavior. This is also reflected by other clinical features of these patients. Patients with ipsilateral PM without MPE as their only evidence of distant metastasis may could to be staged as limited disease. However, our data are limited and potentially biased. More clinical data and the specific mechanism investigation are needed in the future for further elucidation.

Our study classified patients into three groups according to primary cancer location: outer quadrant, inner quadrant, and areolar area/central axis. Pokieser W et al. (47) reported that invasive ductal carcinomas located in the inner quadrants were significantly associated with increased pleural effusion as the first site of metastasis, which may be associated with a higher rate of internal mammary lymph node metastasis. Similarly, our study reported that 46.6% of patients had primary cancers located in the inner quadrant. Furthermore, our results indicated that primary cancer location in the inner quadrant is a poor prognostic factor for patients. Some studies demonstrated that BC situated in inner quadrants have a worse prognosis (48–51), which may cause by the anatomical accessibility of the tumor to the internal mammary lymph node (49, 52). Additionally, growing evidence suggests differences in metastatic spread among BC biologic subtypes (6). Smid M et al. suggested that the majority of pleural relapse occurred in both luminal subtypes (53), which is consistent with our findings. Prognosis of metastatic breast is confirmed to be affected by a combination of factors such as molecular features (54). The prognostic role exerted by pathological factors varies in different disease contexts. Although we observed significant differences in M-OS among the four molecular subtypes of BC ( Supplemental Figures 3A, B ), multivariate Cox results revealed that it was not an independent prognostic factor for patients with only PM. Similarly, Yang Y et al. suggested that the prognosis of patients with cancer with MPE was independent of histology (41). This may be caused by the particularities of the studied patients or by data bias. In addition, BC is highly heterogeneous, and patients with the same molecular subtype also have distinct molecular features, responses to treatment, and prognosis (55, 56). Global burden of molecular mutations into primary tumor and metastatic samples seemed to be independent of the molecular subtype of primary tumor and metastatic sites in the study of Callens C et al. (54). By contrast, one study by Schrijver et al. showed different molecular mutational signatures for different metastatic sites (57). This may be one of the reasons why molecular subtypes did not appear as a predictor of survival in PM patients without other distant metastases, and the mechanisms remain to be further investigated and elucidated. Furthermore, the lymph node status and tumor size were independent predictors of death due to BC (58). Several studies have reported that the higher the T/N stage, the worse the prognosis of patients with BC (59), which was consistent with our results.

Evidence-based guidelines suggest the use of conservative treatments for patients with limited life expectancy, whereas they suggest offering more aggressive treatment modalities for patients with better prognoses. In this study, we focused on analyzing the survival of patients with PM without other distant metastasis at the time of first MBC diagnosis and identifying the prognostic factors. Identifying these characteristics and understanding their prognostic value in diseases could enable customized treatments for this patient group. The nomogram model constructed in this study included all the independent risk factors that we screened, and it provided a visual and user-friendly tool for risk evaluation and prognostic prediction of patients with BC with only PM, facilitating tailored management strategy for these patients.

However, inevitable the study has some limitations. (1) Our study was a single-center retrospective analysis with a limited number of cases, which may have caused some restrictions and biases in the results. (2) Although the nomogram achieved ideal prediction efficacy; it lacked external validation to further enforce the reliability.

BC with PM without additional distant metastasis at the time of first MBC diagnosis exhibited a better prognosis than those with combined LM or LM alone. For patients with PM, the prognosis of patients with MPE was worse. Primary cancer site, T stage, N stage, location of PM, and MPE were identified as independent prognostic factors for predicting M-OS in patients with PM as the only distant metastatic site. The nomogram provided a quantitative method for predicting individual survival in this subset of patients.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

SL, ZY, and CL contributed to the conception and design of the study. SL collected data, performed the statistical analysis, and wrote manuscript. CL, WS, XL, and LS wrote sections of the manuscript. ZY reviewed and revised the manuscript, and acts as guarantor. All authors listed have read the final manuscript and agree to its publication.