Effect of Positive Hepatitis B Surface Antigen on The Risk of Synchronous Liver Metastasis: A Retrospective Study of 868 Consecutive Patients of Newly Diagnosed Gastric Cancer

Background. The study aimed to explore the inuence of hepatitis B virus (HBV) infection on the risk of synchronous gastric cancer liver metastasis (synGCLM). Methods. This was a retrospective study which enrolled 868 patients with newly diagnosed gastric cancer (GC). The study compared the prevalence of synGCLM between hepatitis B surface antigen (HBsAg)-positive (HBsAg + ) and -negative (HBsAg - ) patients. Logistic regression analysis was utilized to analyze the risk factors for synGCLM. Among patients with and without synGCLM, aspartate aminotransferase to platelet ratio index (APRI), liver brosis-4 index (FIB-4) and hepatitis B e antigen (HBeAg) status were further analyzed. Results. The prevalence of synGCLM in the HBsAg + patients was higher than that in the HBsAg - patients, which was statistically signicant (P = 0.025). Multivariate logistic regression analysis demonstrated that HBsAg, the elevated level of carcinoembryonic antigen (CEA), alpha-fetoprotein (AFP), γ-glutamyltransferase (GGT) and alkaline phosphatase (ALP) were risk factors for synGCLM. Among the HBsAg + patients, both ARPI and FIB-4 were signicantly higher in the patients with synGCLM (synGCLM + ) than those without synGCLM (synGCLM - ) (ARPI: P = 0.045; FIB-4: P = 0.047); HBeAg positivity was detected in 20.0% of synGCLM + patients compared to 6.0% of synGCLM - patients, but the difference was of no signicance (P = 0.190). stellate cells; IL, interleukin; IFN, interferon; factor.

Conclusions. HBV infection signi cantly increases the risk of synGCLM, and elevated ARPI and FIB-4 may be pro-metastatic especially among the HBsAg + GC patients.

Background
Gastric cancer (GC) ranks the fth most frequent cancer, and the third most common cause of cancerrelated death over the world [1]. GC is more severe in China as it ranks the second commonly diagnosed cancer in males and the second leading cause of cancer-related death. Newly reported by National Cancer Center, there were approximately 679,100 new cases and 498,000 deaths from GC [2]. GC patients always have a poor prognosis due to presenting with distant metastases at initial diagnosis of the primary tumor.
And the liver is the most common organ of GC hematogenous metastasis as a result of its unique anatomical structure [3]. Previous studies have shown that about 2.0%-9.9% of patients present synchronous gastric cancer liver metastases (synGCLM) at initial diagnosis of primary site, and up to 37% of GC patients develop metachronous liver metastasis after radical gastrectomy [4]. Unfortunately, the 5-year survival rate of synGCLM is less than 20% [5] due to losing the chance of undergoing curative resection and optimal treatments [6][7][8][9][10].
Distant metastasis is one of the most signi cant prognostic factors in advanced GC. The seed and soil hypothesis refers that the compatibility between tumor cells (seed) and a pre-metastatic niche (soil) is needed, which explains the mechanism of cancer metastasis [11,12]. Previous clinical studies focused on the risk factors for gastric cancer liver metastases (GCLM) from the side of 'seed', such as the aggressive behavior of tumor cells [13,14]. However, our article tried to explore the impact of the liverspeci c microenvironment ('soil') on GC metastasis, which rarely debated.
Hepatitis B virus (HBV) infection is prevalent in China as the prevalence of hepatitis B surface antigen (HBsAg) seropositivity is 5%-6% in the general population [15,16], which plays a vital role in the progress of liver diseases including hepatocellular carcinoma (HCC) [17][18][19]. With the notable prevalence of HBV infection in China, we intented to explore the impact of HBV infection on premetastatic microenvironment formation. In the study, by conducting a large-sample, retrospective study of 868 cases with newlydiagnosed GC, we obtained some meaningful and unique results of the impact of HBV infection on synGCLM.

Study population
This is a retrospective, cross-sectional study of all consecutive newly-diagnosed GC patients who presented to the A liated Zhongda Hospital of Southeast University from May 2013 to May 2020. The inclusion criteria were as follows: 1) Newly diagnosis GC by pathological reports of gastroscopic biopsy or surgical specimen in our hospital; 2) No surgery, chemotherapy, radiotherapy or other treatments in other hospitals before admission; 3) Liver metastases con rmed by ultrasound, computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography/computed tomography (PET/CT) or liver biopsy, and complete imaging examination to detect other distant organ metastases; 4) Complete medical records including demographic information, serologic assay, tumor characteristics. We excluded the patients who conformed to the following standards: 1) Pathologically con rmed as gastric lymphoma, gastrointestinal stromal neoplasms, gastric leiomyoma; 2) Accompany with hepatitis virus infection other than HBV or with liver diseases such as fatty liver, alcohol liver and further apparent liver damage; 3) Lack of necessary clinical information; 4)History of primary tumors in other organs except for stomach.
SynGCLM de ned as liver metastasis detected before or at the meantime of diagnosing primary GC, including within six months after the diagnosis of the primary site, since that some suspicious liver lesions needed to follow up to clarify if they were liver metastases or not. HBV infection characterized by HBsAg seropositivity [20]. And HBsAg + was de ned as detection of serum HBsAg more than 10 IU/mL by serologic assay in our hospital.

Statistical analysis
The study compared baseline clinicopathological features between HBsAg + and HBsAg-negative (HBsAg -) patients. Continuous variables with skewed distribution were compared by Wilcoxon rank-sum test and expressed by median (range). Qualitative variables were compared by Chi-square test, Fisher's exact test and described in the form of a percentage. Possible risk factors for synGCLM were analyzed rst by univariate logistic regression analysis, and then performed multivariate logistic regression analysis on the covariates which were of statistical signi cance in univariate analysis. All statistical analyses were performed by SPSS statistical software (version 25.0). And statistical signi cance was considered as two-sided P value < 0.05.

Results
The study enrolled a total of 868 patients with newly diagnosed GC between May 2013 and May 2020 in our hospital. There were 60 (6.9%) patients with HBsAg and 808 (93.1%) without HBsAg. According to the HBV infection status, these cases divided into two groups: HBsAg + and HBsAg -. The clinicopathologic characteristics of these two groups summarized in Table 1. By Chi-square test, the prevalence of synGCLM in the HBsAg + group was 16.7%, which was signi cantly higher than that of 8.2% in the HBsAggroup (P = 0.025). No signi cant difference found in gender, age, tumor location, grade, histological type between HBsAg + and HBsAgpatients. Figure 1 showed the typical CT, MRI and histopathological images of HBV infection leading to liver metastasis in GC patients, respectively.
For patients with synGCLM (synGCLM + ) and without synGCLM (synGCLM -), there were no signi cant differences in terms of gender, tumor location, grade, PLT, TBIL, FIB (P > 0.05) in Table 2. Following univariate logistic regression, HBsAg, CEA, CA199, AFP, AST, ALT, GGT, ALP, histological type and age were statistically signi cant associated with synGCLM, which shown in Table 3. These ten risk factors were further chosen for the subsequent multivariate logistic regression analysis to nd out the independent risk factors of synGCLM ( Table 4). The results demonstrated that not only HBsAg + (vs HBsAg -; OR, 2.849; 95% con dence interval, 1.130-7.186; P = 0.027), but also elevated levels of preoperative CEA, AFP, GGT, ALP were signi cant risk factors for synGCLM. Furthermore, receiver operating characteristic (ROC) curve analysis revealed that the combination of CEA and AFP show perfect sensitivity and speci city for synGCLM in Figure 2.

Discussion
The liver is the most frequent site of hematogenous metastasis in GC, and the patients usually have dismal prognosis when diagnosed with GCLM [21]. Considering that liver metastasis commonly accompanies with simultaneous peritoneal and distant lymph node metastases, the guideline recommends chemotherapy instead of surgery as the rst-line treatment for GCLM [3,22]. The famous seed and soil hypothesis proposes that only the microenvironment that has acquired the capacities of engraftment, survival support and immune evasion could become a new metastatic target organ [23]. In the process of cancer metastasis, cancer cells rst form micrometastases in distant metastatic organs and then proliferate into visible metastases. The speci c organ microenvironment can promote the colonization, survival and proliferation of tumor cells. However, lack of research shed light on the liver microenvironment (soil) in GCLM. To early detect the high-risk individuals of synGCLM, the study intented to declare whether liver disease would remodel its microenvironment to become a fertile, premetastatic soil for GC cells (seed) to land and proliferate. Here we conducted a retrospective study to demonstrate the impact of HBV infection on the metastatic pattern of GC.
We found that the prevalence of HBsAg + was 6.9% among all the GC patients, in accordance with that in the general population in China [16]. The prevalence of synGCLM in our study was 8.8%, which was also consistent with that reported by previous studies [4]. Results illustrated that the prevalence of synGCLM in the HBsAg + patients was 16.7%, signi cantly higher than that of 8.2% in the HBsAgpatients (P = 0.025). As HBV infection affected the metastatic pattern of tumor, a study in Italy reported that the presence of HBsAg in patients with colorectal carcinoma signi cantly increased the incidence of liver metastasis recently [18]. Meanwhile, with the stimulation of HBsAg, interleukin (IL) -10, IL-12 and interferon (IFN) λ could be produced by dendritic cells, which played a vital role in the progression of HBVrelated tumor [24]. Moreover, HBV-encoded X protein (HBx), as the only one protein encoded by the HBV genome and expressed by transformed hepatocytes, could affect differentiation and promote tumor progression metastasis, which proved by the positive cross-talk with the metastasis-associated 1 (MTA1) in stabilizing hypoxia-inducible factor (HIF)-1α [25,26]. Experiments in vivo and vitro also proved that HBx could promote tumor invasion by up-regulating matrix metalloproteinases (MMPs) and cyclooxygenase (COX)-2, which provides new insights into its involvement in tumor metastasis and recurrence [27]. Also, HBx could up-regulate long non-coding RNA MALAT1 and further activate LTBP3, which resulted in the development and metastasis of tumor [28]. Furthermore, the transforming growth factor (TGF)-β production stimulated by HBx could promote Treg cell recruitment, which maintained a tolerogenic liver microenvironment and nally induced the development of metastasis [29,30]. These above mechanisms may explain how HBV infection adjust the liver microenvironment becoming a premetastatic niche (soil) for circulating tumor cells (seed) to colonize. Besides, a study conducted by Wei veri ed the objective evidence of the existence of HBV infection in GC tissues, it proposed that HBV infection was a possible signi cant predictor for the development of GC [31]. The in uence of HBV infection on both GC tumor cells and liver microenvironment may mediate GC metastasis, and the interaction between tumor cells and liver microenvironment need to be further investigated.
Inspired by the classic hepatitis-cirrhosis-HCC pathway [32], we further investigated whether liver brosis/cirrhosis related to HBV infection in uenced GCLM. As liver biopsy was seldom utilized in clinical work, we used the noninvasive liver brosis model to assess the degree of liver brosis in the 60 GC patients infected with HBV. Due to ARPI and FIB-4 were both e cient indexes used to evaluate the degree of HBV-related liver brosis, which took the factors like the age, platelet, AST, ALT of patients into consideration, and widely available in the clinical setting [33]. The study demonstrated that both the median of ARPI and FIB-4 in the HBsAg + patients with synGCLM were signi cantly higher than those among the HBsAg + patients without synGCLM. HBeAg positivity in HBsAg + patients with synGCLM was higher compared with that in HBsAg + patients without synGCLM, although the difference was of no signi cance. A study conducted by Chiou in Taiwan drew a similar conclusion that cancer patients with liver cirrhosis had a higher risk for liver metastasis in colorectal cancer [34]. Another study from China also reported that brosis niche might be a favourable microenvironment for the formation of hepatic metastasis based on noninvasive liver brosis scores [35]. The phenomenon may be resulted from local microenvironment change and immune reaction alteration. The continuous production of HBsAg can cause persistent HBV infection-associated in ammation, which accompanied by the transdifferentiation of hepatic stellate cells (HSC) into myo broblasts. At the meantime, these broblasts will propagate, along with the deposition of the extracellular matrix, consequently lead to liver brosis even cirrhosis [36][37][38]. On the other hand, as the component of prometastatic liver microenvironment, HSC can promote tumor growth, facilitate tumor invasion and suppress the anti-tumor immune response [39]. Moreover, the angiogenic factors such as vascular endothlial growth factor (VEGF) and angiopoietin 1 in liver brosis environment could promote tumor angiogenesis [40]. In conclusion, more animal experiments and prospective clinical trials need to dig into the underlying mechanism of the cirrhotic microenvironment and active virus replication in the metastasis of GC.
There were several limitations in that it was a retrospective, cross-sectional study conducted in only one single center. The comorbidities like diabetes, obesity and the HBV-DNA level of all patients were insu cient. And liver brosis/cirrhosis was measured by ARPI/FIB-4, other than a liver biopsy. Despite these limitations, our study still enrolled a large number of newly diagnosed GC patients, and all with information about hepatitis B status. Moreover, our study investigated the effect of HBV infection only on synchronous liver metastasis, to avoid the interference of different anti-cancer treatments among GC patients and imprecision information about metachronous metastasis. Besides, our study was the rst study that demonstrated the impact of not only HBV infection but also HBV-related cirrhosis on liver metastasis formation and survival in GC. Overall, our study discovered that HBV infection and HBVrelated cirrhosis might remodel the liver microenvironment, becoming a premetastatic niche for GC cells to proliferate. This unique and exciting information may provide novel anti-cancer treatments on blocking premetastatic niche formation (soil) to prevent liver metastasis and even cure gastric cancer metastasis in the future. More prospective studies with larger sample size and animal experiments are needed to validate our ndings and further explore the cause-effect relationship between HBV infection and synGCLM.

Conclusion
Our study found that HBV infection may increase the risk of liver metastasis in GC patients, and for hepatitis B surface antigen positive (HBsAg + ) GC patients, HBV-related liver brosis and cirrhosis may be prometastatic. This may provide novel anti-cancer treatments on blocking premetastatic niche formation to prevent liver metastasis. More investigations need to be taken to explore the mechanism of how HBVrelated microenvironment promoting liver metastasis.
Authors' contributions TYW complete data collection, data analysis, manuscript drafting. HJZ directed the arrangement of the study and supervised the whole writing of the manuscript and manuscript revision. All the authors contributed to the review and revison of the manuscript and have read and revised the nal version.

Funding
Not applicable.

Availabilty of data
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate
The study was approved by the ethical committee of the A liated Zhongda Hospital of Southeast University.

Consent for publication
Not applicable.

Competing interests
The authors declare that they have no competing interests.
Author details      Figure 1 CT, MRI and histopathological images of HBV infection leading to liver metastasis in GC patients (a) The liver had a regular shape, and a mass of low density shadow with ring-like enhancement could be seen in the right lobe of the liver, which had an unclear boundary. A large amount of liquid density shadow was seen in the abdominal cavity. (b) Multiple mass and nodular abnormal signals were seen in the liver parenchyma with progressive ring-like enhancement, and the diffusion was obviously limited. Massive ascites could also be seen around the liver. (c) Magni ed 100 times, hematoxylin-eosin staining: Fibrotic changes and in ammatory cell in ltration were observed in tumor-adjacent tissues. There were punctate and focal necrosis of cancer cells, yperplasia of connective tissue in metastases. The liver metastasis was in accordance with the origin of gastric cancer.

Figure 2
Receiver operating characteristic analysis for the prediction of synchronous liver metastasis in gastric cancer patients