Distribution of primary site of hepatic metastases and prognosis of hepatic metastases at different primary sites

Background Most of the time, the primary tumor of hepatic metastases is easy to find. However, sometimes this is a confusing question. The present study aimed to explore the distribution of primary sites of hepatic metastases and the prognosis of hepatic metastases at different primary sites. Methods Data on patients with hepatic metastases were obtained from the surveillance, epidemiology, and end results (SEER) database between 2010 and 2016. Descriptive statistics was used to analyze the distribution of primary sites, life table to calculate survival rate, Kaplan-Meier method to plot survival curves, and Cox univariate regression and multivariate regression analysis to assess prognostic factors. Results The median age of 151,821 patients was 67 years; 58.7% of the primary sites were derived from the digestive system, followed by the respiratory system (25.6%) and other sites (15.8%). Ten leading primary sites were lung (25.4%), colorectum (24.1%), pancreas (19.8%), stomach (4.5%), breast (4.1%), gallbladder and biliary system (3%), esophagus (2.7%), kidney and renal pelvis (2.1%), ovary (1.8%) and primary sites unknown (2%). The overall survival probability of patients with hepatic metastases at 6 months, 1 year, and 3 years was 44.5%, 28.7%, and 9.9%, respectively, with a median survival of 4 months. The prognosis of hepatic metastases from lymphoma (Hodgkin and non-Hodgkin), the small intestine, breast, and colorectum was relatively better, and that from primary sites unknown, urinary system (bladder, urethra), lung, and pancreas was quite poor. Conclusion The distribution of primary sites of hepatic metastases was different under different grouping conditions, but the lung, colorectum and pancreas were the top three sources. The prognosis of hepatic metastases from different primary sites varied greatly.

colorectum (24.1%), pancreas (19.8%), stomach (4.5%), breast (4.1%), gallbladder and biliary system (3%), esophagus (2.7%), kidney and renal pelvis (2.1%), ovary (1.8%) and primary sites unknown (2%). The overall survival probability of patients with hepatic metastases at 6 months, 1 year, and 3 years was 44.5%, 28.7%, and 9.9%, respectively, with a median survival of 4 months. The prognosis of hepatic metastases from lymphoma (Hodgkin and non-Hodgkin), the small intestine, breast, and colorectum was relatively better, and that from primary sites unknown, urinary system (bladder, urethra), lung, and pancreas was quite poor. Conclusion The distribution of primary sites of hepatic metastases was different under different grouping conditions, but the lung, colorectum and pancreas were the top three sources. The prognosis of hepatic metastases from different primary sites varied greatly. Background 3 The primary site of the tumor determines its biological behavior and metastatic characteristics [1][2][3], and different tumors have completely different morbidity and mortality [4]. According to the conventional strategy, doctors need to develop individualized treatment plans based on factors, such as the primary site, metastasis, and comorbidities of a tumor. However, in the actual clinical process, many tumors are found not because of the symptoms of the primary site, but the symptoms of metastases or accidentally found metastases, rather than primary sites during routine physical examination [5]. With the increasing accuracy of ultrasound, computed tomography, and magnetic resonance imaging, especially contrast-enhanced examinations, even small hepatic tumors are easily detected during routine examinations [6][7][8].
Due to the generally poor prognosis of patients with hepatic metastases, it is critical to quickly identify the primary site of tumor and promptly treat it. The liver is the most predominant metastatic organ of various tumors, such as pancreatic cancer [9], colorectal cancer [10], and stomach cancer [11]. According to this characteristic, when hepatic metastasis is suspected, we often check primary lesions in these organs. Most of the time, the primary tumor is easy to find. However, sometimes this is a confusing question.
Although positron emission tomography-computed tomography or systemic diffusionweighted imaging [12] can be of great help in finding metastases or primary lesions, doctors still have difficulty in exploring the primary site.
So far, by consulting the literature, we have not found a study on the distribution characteristics of the primary site of hepatic metastasis. Therefore, we extracted the clinical characteristics of patients with hepatic metastasis from the surveillance, epidemiology, and end results (SEER) database, with the purpose of analyzing the distribution of the primary site and the survival status of each tumor type with hepatic metastasis; thus, providing guidance to help physicians to design imaging or other examinations to find the primary site, and to make decisions regarding curative-intent interventions.

Data
Data were derived from the SEER database which started to release metastatic information related to liver, lung, bone, and brain in 2010. From the latest version of April 2019, the most recent follow-up cutoff date was December 31, 2016. The colorectum was defined as originating from the cecum and ending at the rectum, including the entire large intestine, with ICD-O-3 coded as C180-C209, C260. Other digestive systems were located in the gastrointestinal tract but not specifically recorded (C268, C269, C488). In the primary site classified by system, "others" refers to all systems except the digestive system and respiratory system (excluding myeloma and leukemia).
Surgery of primary site was defined when the primary site received surgical interventions, regardless of whether it was partial or radical resection, or whether the metastases which were treated was unknown. Combined metastases was defined when hepatic metastasis was accompanied by at least one or more of lung metastasis, bone metastasis, and brain metastasis. Age at diagnosis was categorized as younger than 45 years, 45 to 64 years, 5 and 65 years and older.

Statistical Analysis
Cox regression analysis and descriptive statistics of patients' demographic and tumor characteristics were summarized by SPSS 25 (IBM Corp., Armonk, NY). Survival probability and survival curves were calculated and plotted using the R "survival" and "survminer" package; the pie chart and forest plot were also drawn using R (https://www.rproject.org/). Statistical significance was set at two-sided P < 0.05. Table 1 shows demographic data and clinical baseline characteristics of a total of 151821 patients with hepatic metastases (80352 men and 71469 women, 1.12:1) were investigated. The median age was 67 years (IQR 58-76) and the median follow-up time was 3 months (IQR 1-11). Among them, 58.7% of the primary sites were derived from the digestive system, followed by the respiratory system (25.6%) and other sites (15.8%).

Patients' characteristics
Also, 80.4% of patients did not receive surgical-related interventions and 41.2% of patients had metastases in other sites, such as the lungs, bones, or brain. Other specific information is shown in Table 1. Table 2 shows number and percentage of the specific distribution of the primary site, in which there were no patients with hepatic metastasis in myeloma and leukemia. At the same time, we conducted a subgroup analysis based on sex and age. Ten leading primary sites for the hepatic metastasis by sex and age were arranged in order, as shown in Figure   1.
In addition, the esophagus and kidney were more common in men than the gallbladder and biliary tract, but this trend was opposite in women.
In patients aged 65 years or older, the common primary sites were the lung and bronchus (28.5%), pancreas (21.6%), and colorectum (21.2%) ( Figure 1F). With increasing age, the proportion of primary sites of colorectal, breast, and genital system decreased, but the proportion of pancreas, lung and bronchus, gallbladder, and biliary system gradually increased. Table 3 shows the 1-year OS of patients with hepatic metastases at each primary site. The overall survival rates of patients with hepatic metastases at 6 months, 1 year, and 3 years 7 were 44.5%, 28.7%, and 9.9%, respectively, with a median survival time of 4 months ( Figure 2A). The prognosis of hepatic metastases from lymphoma (Hodgkin and non-Hodgkin), small intestine, genital system (breast, ovary, and other male or female genital organs), retroperitoneal tissues, and colorectum and anus was relatively better. The prognosis of hepatic metastases from primary sites unknown, urinary system (bladder, urethra), lung, pancreas, hepatobiliary system, and esophagus was quite poor (Supplementary Figure 1). Table 4 shows the risk factors for the prognosis of hepatic metastases. Through Cox univariate and multivariate analysis, we found that sex, age, marital status, ethnicity, primary site, grade, presence of surgery, and presence of combined metastases were independent risk factors for prognosis. Figure 2 shows the Kaplan-Meier survival curves for patients with hepatic metastases under different conditions. Supplementary Figure 2 shows a forest plot based on univariate Cox regression analysis of the effects of surgery on the prognosis of each primary site. Most of hepatic metastases benefitted from surgical intervention at the primary site, except for the anus, non-Hodgkin lymphoma, prostate, eye and orbit, and other nonepithelial skin except melanoma of the skin.

Discussion
In this study, we depicted the distribution of primary sites of hepatic metastases and demonstrated that there were significant differences in the distribution of primary sites of hepatic metastases under different conditions. The prognosis of hepatic metastases from different primary sites varied greatly. Surgical intervention at most of the primary sites can benefit the patient's survival with hepatic metastases.
Metastasis is the process that includes local invasion by the primary tumor cells, intravasation into the blood (hematogenous) and/or lymphatic system (lymphatic), arrest at a distant organ, and metastatic colonization [13].It is also the process by which a localized cancer becomes a systemic disease [14] and the cause of about 90% of cancerassociated deaths [15]. The preferred sites of metastasis for a given type of cancer often include the first capillary beds downstream of the primary tumor [15]. The liver is rich in blood and is supported by both the hepatic artery and the portal vein. Therefore, it becomes a target organ for hematogenous metastasis and is threatened by all malignant tumors that produce circulating tumor cells (CTCs).
CTCs originating from the lungs can enter the liver through the hepatic artery, so the liver is the target organ of lung cancer. There were approximately 228,150 estimated new lung cancers in 2019, far more than other organs [4]. Therefore, although the proportion of lung cancer metastasized to the liver in the four most common target organs of lung, bone, brain, and liver is the least, this value is about 16%-17% [16]. However, due to the large number of the patients suffering from lung cancer, 49.3% of whom have had distant metastases at the time of diagnosis [17], the absolute sum of patients with hepatic metastasis is tremendous. In the present study, when sorted by organs, tumors originating from the lung accounted for 25.6% of the total, ranking the first.
The digestive system delivers blood to the liver through the portal system. When a digestive tract tumor exists, the portal circulation, the most common route of dissemination of the digestive system tumor, contains both nutrients and CTCs that cause hepatic metastasis [18]. Therefore, the liver becomes the most preferred target organ for digestive system tumors, such as those of the colorectum [10], pancreas [9], stomach [11], and extrahepatic bile-duct [19]. In the present study, primary tumors derived from the digestive system accounted for 58.7% of the total (Table 1); the predominant sites in the system were the colorectal, pancreas, stomach, and gallbladder and other biliary tumors.
The influence of sex on the origin of the primary site was evident, and this difference 9 could be due to the demographic and clinical characteristics of the primary tumor.
Esophageal cancer [20], stomach cancer [21], and kidney cancer [22] are more common in men than in women, especially esophageal cancer (about 3:1). In men, the proportion of these three tumors is higher than that in women. In contrast, gallbladder and biliary tract tumors are more common in women [23], so hepatic metastases from the gallbladder and other biliary tumors in women are more common than that in men. In addition, the sum proportion of metastases in female-specific sexual organs, such as the breast, ovary, and uterus are relatively large (17.9%). As it is a specific feature of women, we should pay attention to it for women with hepatic metastases. Although morbidity of prostate cancer is the first one in men, primary tumor originating from the organ is rare because only about 7.53% of patients have metastases at diagnosis and the liver is not the preferred metastatic site for it [24].
The origins of the primary sites of different age groups show different characteristics, and the reasons for these differences may be the same as sex. At present, we have not found any research that can provide a perfect explanation of this discovery. We speculate that the reason colorectal, breast, pancreatic, and genital system-related tumors occupy a large proportion in young patients is that the number of colorectal cancer or breast cancer patients is large [4], and the liver is the preferred metastatic organ of colorectal cancer [10] and pancreatic cancer [9]. Furthermore, pancreatic cancer has a very high rate of hepatic metastasis [9], and the median age of breast cancer [25] and genital systemrelated tumors is relatively low (less than 65 years) [26,27].
The difference in prognosis of hepatic metastases originating from different sites, with 1year OS as an example, from 12.2% of the primary site unknown to 78.7% of Hodgkin's lymphoma, is very significant (Table 3 and Supplementary Figure 1). Multivariate analysis found that sex, age, marital status, ethnicity, primary site, grade, presence of surgery, and presence of combined metastases were independent risk factors for prognosis. The reason for the poor prognosis of patients with unidentified primary sites is that if the primary tumor is not found, we cannot provide special treatments, such as surgery, radiotherapy, or chemotherapy. In contrast, chemotherapeutics of Hodgkin's lymphoma has changed its prognosis from being relatively incurable to one in which patients have a high likelihood of long-term survival [28].
This study found that surgical intervention at most of the primary sites can benefit patient's survival with hepatic metastases. In patients with a better prognosis as described in the previous paragraph, the primary site received a higher rate of surgical intervention (Supplementary Figure 2). Other studies have found that hepatectomy for synchronous gastric cancer hepatic metastases may carry survival benefits in selected patients [29]. In patients with colorectal hepatic metastases, surgical resection represents the only chance of long-term survival [30]. In patients with pancreatic cancer, hepatectomy is a procedure with a potential survival benefit for carefully selected Although this study systematically described the origin of the primary site of hepatic metastases and the survival status of patients with hepatic metastases, there are still some limitations. First, the dataset used for this study did not include data regarding number, size, anatomical location, or distribution of hepatic metastases, which are important factors that can affect the survival results presented in this study [29]. Second, it was limited by the dependence on hospital administration, such as the lack of agreement between workers on diagnosis codes, which could lead to registration errors [34].

Conclusions
This study systematically described the distribution of the primary site of hepatic metastases and performed a subgroup analysis based on sex and age. From the results, we conclude that there were significant differences in the distribution of primary sites of hepatic metastases under different conditions. In addition, this study also evaluated the prognosis of patients with hepatic metastases from different primary sites. We believe that this work will have a positive impact on guiding clinicians to find the primary site of hepatic metastases and formulating treatment strategies.

Supplementary Files
This is a list of supplementary files associated with the primary manuscript. Click to download.