Heterogeneity of Synchronous Lung Metastasis Calls for Risk Stratification and Prognostic Classification: Evidence from a Population-Based Database

Simple Summary Cancer patients with synchronous lung metastasis (sLM) are recognized as an entity with poor survival, and no consensus exists about which patients may benefit from active treatment. The current study demonstrates disparities in the prevalence and prognosis of sLM by primary cancer type and clinicodemographic factors, based on the SEER database. These heterogeneities here lay a foundation for and call for the development of risk assessment and prognosis classification tools that drive clinical management. Abstract The epidemiology and associated potential heterogeneity of synchronous lung metastasis (sLM) have not been reported at a population-based level. Cancer patients with valid information about sLM status in the Surveillance, Epidemiology, and End Results database were enrolled. The prevalence of sLM, with a 95% confidential interval, and median survival of sLM, with interquartile range, were calculated and compared by Chi-square analyses and log-rank tests by primary cancer type and clinicopathological factors. Furthermore, the risk factors of sLM development were identified by multivariate logistic regression. Among 1,672,265 enrolled cases, 3.3% cases were identified with sLM, with a median survival of 7 months. Heterogeneity in prevalence and prognosis in sLM was observed among different primary cancers, with the highest prevalence in main bronchus cancer and best survival in testis cancer. Higher prevalence and poorer prognosis were observed in the older population, male population, African American, patients with lower socioeconomic status, and cases with advanced T stage, N stage, or more malignant pathological characteristics. Race, age, T stage, N stage, metastasis to other sites, insurance status and marital status were associated with sLM development (p < 0.001). The current study highlights the heterogeneity of the prevalence and prognosis in patients with sLM.


Introduction
Cancer represents a great threat to public health worldwide, with an estimated 18.1 million new cases and 9.6 million cancer-related deaths globally in 2018 [1]. Approximately 90% of cancer-related deaths can be attributed to cancer metastasis, a major indication of treatment failure [2]. Early detection and active treatment of primary lesions may help reduce the risk of the development of metastatic disease [3,4]. The prognosis for metastasized cases is generally poor due to high tumor burden, inferior performance status, limited therapeutic options, and impairment of the involved organs' function. The lungs are the third most common metastatic site for cancer, following the liver and bone [5]. Based on their organ mechanics, specifically vein drainage from the systemic circulation and their large area of micro-circulation, the lungs are exposed to the great risk of metastasis. Additionally, the organotropism is also attributable to the exosomes secreted by the primary cancer cells to prepare the premetastatic niche [6].
Great heterogeneity among metastatic lesions concerning tumor burden, treatment response, prognosis, and so on challenges the crude "yes-or-no" classification system for stratifying the metastasis status of cancer patients. For instance, a modified M stage by the organs being metastasized was proposed in pancreatic neuroendocrine tumor [7]. In another perspective of metastasis, extra heterogeneity also exists in terms of metastatic origin; that is, patients with metastatic lesions from different primary cancers show divergent disease characteristics, treatment responses, prognoses, and so on [8]. Cancer metastasis was conventionally believed to be incurable; however, such heterogeneities raise hope for selecting patients who may potentially benefit from the evolving therapeutics from palliative or best supportive cares towards active interventions or even curative attempts. Individualized recommendation of treatment protocol is based on the precise prognosis prediction and thus the selection of patients who are expected to benefit more from the treatment. For sLM, currently available treatment includes stereotactic body radiation therapy, surgery, and ablation, followed by an in-depth understanding of cancer biology [9]. However, the heterogeneity of seeding primary cancer cells which also drive the prognosis disparity remains largely unknown and the few studies are elaborating on prognostic stratification of patients with sLM. There are previous publications on relevant topics that mainly focus on the molecular mechanism of sLM formation, retrospective studies on the treatment of sLM, and epidemiological studies based on small sample size, or from only colorectal cancer or breast cancer [10,11]. In addition to the heterogeneity of metastasis caused by biological factors, the formation of metastasis may be also influenced by socioeconomic status (SES) [12].
The current study was designed to test the hypothetical disparities with regards to the prevalence and prognosis within sLM patients with different biological features of the primary tumors and different socioeconomic statuses across different cancer types or even sub-types within certain cancers. To achieve sufficient representativeness and generalizability, we explored the Surveillance, Epidemiology, and End Results (SEER) database to achieve estimations of the prevalence and prognosis of sLM. Point estimates of the prevalence along with 95% confidence interval and median survival, along with interquartile range, were used to compare risk of developing sLM and prognosis.

Eligible Patients
The SEER database, founded by the National Cancer Institute in 1973, currently enrolls cancer cases from 18 registries, covering about 28% of all the U.S. population, and an increasing number of variables were collected by the SEER database over time [13]. The metastatic status of the lungs was first available in 2010; thus, only cases between 1 January 2010 and 31 December 2016 (most updated) with valid synchronous lung metastasis (sLM) status were eligible for this study. sLM is most likely diagnosed by radiology exams, which, however, failed to discriminate multi-focal primary lung cancer (all nodules are primary lung cancer) from metastatic lung cancer (nodules are primary and metastatic). Further pathological and bioinformatic results for these nodules are currently unavailable for clear classification in the SEER database. Thus, to ensure the quality of the current study, the cases of primary lung cancer anatomically in the lungs were excluded. Additionally, bronchus cancer was enrolled to increase the generalizability of this study, since bronchus can be easily discriminated from primary lung cancer in lung tissue by radiology exams. Cases with a prior cancer history or with T stages such as T0, Tis, or Ta (for bladder cancer) were excluded. Only sLM cases with the diagnosis of their primary cancers by either histological or cytological exams were enrolled, and leukemia and lymphoma cases were excluded due to their diffuse nature.

Tumor Classification and Statistical Analyses
The primary cancer types were determined by the information of both organ site and histological diagnosis, except for sarcoma and melanoma, which are defined solely by histology [14,15]. Embryonal tumors include medulloblastoma, Wilms tumor, and neuroblastoma. The subgrouping of cancer types was determined for any particular cancers based on detailed clinical information. Age at diagnosis was categorized into five stratifications (0-18, 19-40, 41-60, 61-80 and 81+ years). The race here includes four classes: Caucasian, African American, other races, and unknown, based on the SEER database [16]. The patients are categorized by the AJCC 7th TNM stages [17].
The prevalence of sLM along with its 95% confidence interval and the ratio of sLM over all metastases were calculated, following the summary of the number of all cases, metastatic cases, and sLM cases by primary cancer (Tables 1 and 2) [18]. The prognosis for sLM cases was depicted by median survival and interquartile range, estimated by the Kaplan-Meier method, and the survival difference was assessed by log-rank tests. Comparison by biological variables and socioeconomic factors was also performed. Here, due to the unavailability of individual-level income, and education information, county-level surrogates were adopted by linking the residing address to the 2013 Rural-Urban Continuum Codes from the United States Department of Agriculture [19]. Multivariate logistic regression adjusted for primary cancer site was performed to assess the relationship between clinical demographic factors and the occurrence of sLM patients. Statistical analyses were performed on R 3.6.0 (https://www.R-project.org/ (accessed on 8 January 2022)), with the survminer and survival package [20,21].
sLM is the major type of metastasis for metastatic cases originating from bone (70.57%), thyroid (63.28%), testis (63.04%), soft tissue (61.66%), kidney (60.88%), and head and neck (51.12%) ( Table 1). In terms of the distribution of primary cancers, 17.48% of sLM cases originated from CRC, with 10.82% from the breast, 10.55% from the kidney, and 10.29% from the pancreas (Table 1). Demographic factors (race, age), clinical factors (T stage, N stage, bone metastasis, brain metastasis, liver metastasis), and socioeconomic factors (marital status, insurance status) are associated with the development of sLM, as revealed by multivariate logistic regression adjusted by primary cancer site (Table 3).

Survival Analysis
The median survival and corresponding interquartile range for sLM cases by primary cancers are presented in Table 1. The median survival for general sLM cases is 7 months, with the best survival in testis cancer and embryonal tumors (median survival unreached), prostate cancer (22 months), breast cancer (20 months), bone tumor (19 months), ovary cancer (16 months), thyroid cancer (11 months), and CRC (10 months) (Table 1). Similarly, survival disparities can be observed in patients with sLM among different age groups, sexes, races, different T or N stages, and patients with different socioeconomic statuses (insurance, marriage, income, residence type, education, and unemployment) ( Table 2,  Supplementary Tables S1-S5). Better prognosis in patients with sLM was observed in females, other race, younger patients, patients who resided in richer or better-educated counties, metro or urban cities, insured patients, and married patients (Figures 2 and 3). Consistent with the counterintuitive prevalence pattern in gastrointestinal cancers, a better prognosis was observed in the slightly advanced cases of these cases with sLM.

Discussion
The current study has reported for the first time the prevalence and prognosis of synchronous lung metastasis (sLM) by different primary cancer types, based on a populationbased database with great generalizability and representativeness. Furthermore, subgroup analyses confirm hypothetical disparities in prevalence and prognosis inside subgroups by both biological factors and socioeconomic factors. These disparities here may highlight the heterogeneity of sLM, which calls for the development of clinical tools for risk stratification and prognosis classification.
The diverse prevalence and prognosis observed here may be biologically attributed to two aspects: microenvironment in lung tissue and primary cancer cell. The "soil," or microenvironment, of lungs influences the formation of sLM, with smoking being verified as a risk factor by clinical retrospective studies and bench studies [22]. Similarly, a retrospective study of 567 pathological stage I, II, or III colorectal cancers showed that current smokers harbor a higher risk of sLM (hazard ratio = 2.72, 95% CI 1.18-6.25; p = 0.02) [23].
Smoking can promote the formation of sLM in colorectal cancer animal models [24]. In the B16-MO5 melanoma animal model, smoking exposure impairs the NK cell-dependent anti-tumor immunity, which can be resecured by activation through bone marrow-derived dendritic cells [25]. A case-control study with 87 female breast cancer patients with sLM and 174 controlled cases without sLM showed a relationship between active smoking and the development of sLM, which was further validated by an animal study [26,27]. Smokeassociated chronic inflammation in the lung microenvironment may promote growth of metastatic cancer cells from breast cancer in mice experiments and patients [28,29]. Interestingly, e-cigarettes also promote breast cancer cell growth and metastatic lung colonization, mediated by cross-talk with tumor-associated macrophages via CCL5 and VCAM-1 pathways in an animal study [30]. However, due to a lack of disease information about the intrinsic lung disease in the SEER database, we failed to verify these findings. Additionally, the development of clinical management of lung metastasis, including surgical techniques and post-treatment care, may help improve the prognosis and quality of life [31,32].
Besides the soil, the pattern of metastasis epitomizes the role of heterogeneous cancer biology. In the subgroup analyses, heterogeneities in tumor biology (tumor marker level, histology characteristics) are translated into disparities in prevalence and prognosis, indicating that involvement of a detailed molecular profile may boost more precise estimation, such as what we achieved in the prognostic classification of brain metastasis. Elevated tumor biomarker level, including AFP, CEA, PSA, and CA125, is usually associated with either higher tumor burden or increased malignancy and thus the intuitively higher prevalence of sLM and poorer survival. HER+ breast cancer is associated with a higher prevalence of sLM and liver metastasis in clinic, which is mediated by its switching effect of TGFβ from inhibiting cell proliferation to promoting cell migration [8,33]. Nottingham grade was believed to be predictive in cancer-specific survival and disease-free survival in breast cancer, and we also observed its predictive role in breast cancer sLM cases, with lower prevalence and better survival in patients with a lower Nottingham grade [34]. RLIP is associated with breast cancer growth and metastasis to the lungs, based on results from an animal study [35]. In line with a bench study, AFP is associated with an increased risk of developing sLM by up-regulation of metastasis-related proteins [36]. The disparity in the prevalence of sLM by fibrosis status can be partly explained by their baseline characteristics: compared with cirrhotic HCC cases, patients without cirrhosis generally present worse pathological grade, larger tumor, and more lymph node involvement [37]. Interestingly, sidedness of CRC also impacts the formation and prognosis of sLM, with lower prevalence but poorer prognosis in right CRC, which may be explained by a more extensive metastasis pattern, poorer differentiation, and a higher percentage of KRAS and/or BRAF mutation in right CRC [38]. Unlike the divergent role of RAS in colorectal cancer liver metastasis, RAS mutation is associated with a higher prevalence of metastasis in lung, brain, and bone, with hazard ratios of 1.5, 3.7, and 1.6, respectively [39,40]. Interestingly, we discovered a counterintuitively higher prevalence of sLM in the patients with T1 stage than in the cases with more advanced stages in esophagus cancer, gastric cancer, and CRC. This disproportional relationship also exists in the prevalence of liver metastasis and brain metastasis in most of the cancer types mentioned above [8,41]. A bioinformatics study in CRC identified a set of genetic mutations associated with the development of metastasis years before a clinically detectable primary lesion [42]. The epidemiology data here may represent clues for bench researchers, who may decode the molecular interplay underlying the clinical manifestations.
SES not only affects primary cancer diagnosis, treatment, and prognosis but modifies the risk of developing metastasis and alters its prognosis [43]. The higher sLM prevalence and poorer sLM prognosis in patients with lower SES status can be explained by the lower awareness of early diagnosis, delayed treatment, and poorer access to healthcare resources [44]. For example, married patients are associated with better survival due to a higher rate of surgery or radiotherapy in all cancer patients as well as sLM patients [45]. Furthermore, the first impression of lower SES in unmarried patients may lead to less intensive treatment from clinicians [46]. Lower income was reported to halt early diagnosis and sufficient and timely treatment, which can be restored by insurance [47][48][49][50][51]. In terms of racial disparity, a higher risk of sLM in African Americans and other races can be attributed to the lower insurance coverage in the population [52], advanced stage of primary cancer at diagnosis [53], less participation in cancer screening [54], and so on. These factors interplay with each other and contribute to the disparities here, and there is a pressing need to increase awareness of this disparity.
Besides highlighting the disparities of sLM, the current study also provides descriptive data about sLM epidemiology. The prevalence data here help estimate the disease burden for pharmaceutical companies, healthcare insurance systems, allocation of lung-directed treatment resources, and so on. As most of the current treatments are designed to target the characteristics of tumor of origin, rather than the metastatic niche, non-metastatic patients are the most frequently enrolled cases and the prevalence data here may help estimate how many patients are needed during initial enrollment, and more importantly, this estimation can be more precise after considering the subgroup disparities. Also, the survival data here may help estimate the sample size for clinical trials for a given expected survival difference.
Despite novel findings in the current study, it harbors the following limitations. Firstly, the SEER database only provides qualitative information about the sLM status, which disenables quantitative analyses by tumor number and tumor size. Secondly, the diagnostic method for the sLM remained unreported, and the heterogeneity in the detecting sensitivity may have biased the prevalence data here. Thirdly, the co-existing lung disease status and the detailed genetic information of primary cancer (such as RAS mutation status in CRC, BRAF mutation in melanoma, and so on) remained unknown, which disenabled more precise epidemiological data on sLM.

Conclusions
The current study reports the prevalence and prognosis of sLM in a pan-cancer manner based on a population-based database with great representativeness. These data may (1) demonstrate the heterogeneity of sLM by providing epidemiological evidence and (2) provide descriptive epidemiology data for estimating sLM burden for the pharmaceutic industry and policymakers, decision-making concerning lung surveillance or screening, and designing clinical trials by providing data to balance the prevalence and survival disparities.   Institutional Review Board Statement: Ethical review and approval were waived for this study due to the fact that the data are fully de-identified and no intervention on patients was performed.
Informed Consent Statement: Patient consent was waived due to the fact that the data are fully de-identified and no intervention on patients was performed.
Data Availability Statement: All data here are publicly available in the SEER database (https://seer. cancer.gov/ (accessed on 12 December 2021)).