HBV Infection Statuses Indicate Different Risks of Synchronous and Metastasis Liver Metastasis in Colorectal Cancer

Background: Previous studies on the effect of HBV infection on CRLM are contrary. This study identied more specic and different impacts of HBV on CRLM. Methods: A total of 3132 CRC patients were analyzed retrospectively and followed up for ve years. All patients were divided into three groups: group A, HBsAg positive with HBV infection; group B, HBsAg negative with HBV infection; and group C, no HBV infection. The risk factors for SYN-CRLM, MET-CRLM, 5-year OS, and LDFS were analyzed. Results: A total of 829 patients had CRLM. SYN-CRLM was found in 425 patients. The incidence of SYN-CRLM was 16.95%, 11.60% and 13.50% (P <0.03) in groups A, B, and C, respectively. HBsAg-positive HBV infection increased the risk of SYN-CRLM (P<0.01), with a worse prognosis in group A (P<0.05). MET-CRLM was found in 404 patients. The incidence of MET-CRLM in groups A, B, and C, was 16.51%, 11.53% and 16.50% (P=0.02). HBsAg-negative HBV infection decreased the incidence of MET-CRLM (P=0.02) with a better 5-year LDFS (P=0.01), but was not related to 5-year OS (P=0.15). Conclusion: HBsAg positivity infection increased the risk of SYN-CRLM with poor prognosis. HBsAg-negative infection reduced the risk of MET-CRLM with better LDFS after surgery.


Introduction
In recent years, the global incidence of colorectal cancer (CRC) has continued to rise. CRC is the third most common malignant tumor with the second-highest mortality rate 1 . In China, the incidence and mortality of CRC are ranked fourth and fth 2 . The majority of CRC patients who died had distant metastasis, and liver metastasis was the most common. Synchronous CRC liver metastasis (SYN-CRLM) was shown in 14-20% of CRC patients at the time of diagnosis [3][4][5] . In patients without SYN-CRLM, nearly 17% of them suffered from metachronous CRC liver metastasis (MET-CRLM), even after resection of the primary tumor 6,7 . According to statistics, the SYN-CRLM rate of colorectal cancer in China is about 25-30%, and the MET-CRLM rate is about 12-21%.
Hepatitis B virus (HBV) infection is the most common chronic liver disease in humans. It is estimated that approximately 2 billion people worldwide have been infected with HBV, and more than 350 million people have chronic hepatitis B (CHB) 8 . Approximately 120 million people in China are chronic HBV carriers, accounting for approximately one-third of the world's total population, and 30 million of them have HBV virus replication throughout their life 9,10 . Hepatitis B is one of the causes of hepatocellular carcinoma, increasing the risk of intrahepatic metastasis [11][12][13] . In addition, it has been reported that HBV infection also increases the risk of liver metastasis in pancreatic cancer, B-cell lymphoma, and other malignant tumors [14][15][16] .
Several previous studies have reported the risk of HBV for CRC liver metastasis, but there are two different opinions. Utsunomiya et al. suggested that the risk of CRLM is lower in patients with HBV and HCV infections 17 . Song et al. suggested that HBV replication can reduce the incidence of CRLM 18   There are variable de nitions of synchronous and metachronous CRLM reported in the literature. According to the statements from the Expert Group on OncoSurgery management of Liver Metastases group (EGOSLIM), SYN-CRLM should be termed "synchronously detected liver metastases" 23 . We diagnosed the intrahepatic nodule as SYN-CRLM at the rst diagnosis of CRC and MET-CRLM at any time after that.
In this study, for the intrahepatic nodule, after the rst diagnosis of CRC, we examined the patient's radiology images, detected the serum CEA and alpha-fetoprotein (AFP) levels, and performed ultrasoundguided biopsy and pathological diagnosis when necessary. After excluding primary liver cancer, hemangioma, and liver cyst, all the results were judged by two radiologists who had more than 5 years of experience in speci cally diagnosing CRLM.

Judgment of HBV infection
All patients were tested for HBV using HBVM ( ve items of hepatitis B) at the rst diagnosis of CRC. According to the results of HBsAg, hepatitis B surface antibody (anti-HBs), hepatitis B e antigen (HBeAg), hepatitis B e antibody (anti-HBe), and hepatitis B core antibody (anti-HBc). We divided patients into three groups: groups A, B, and C. Group A included patients with HBsAg-positive HBV infection. Group B included patients with HBsAg-negative HBV infection who were positive for HBeAg, anti-HBe, and anti-HBc. Group C included patients who were negative for HBVM or were only anti-HBs-positive.

Follow-up
The time between the rst diagnosis of CRC and the diagnosis of MET-CRLM was de ned as LDFS, and the time between the rst diagnosis of CRC and death was de ned as OS. CT or B-ultrasound examinations were performed every 6-12 months after the operation. If abnormal nodules in the liver were found, the diagnosis of CRLM was made by contrast-enhanced ultrasonography or MRI. If necessary, a biopsy was performed for pathological diagnosis. The follow-up o ce of the Sixth A liated Hospital of Sun Yat-sen University followed up all the patients. We collected and recorded the follow-up data through the patient's return visit. For the patients who are examined in a local hospital due to living in a remote area, if there was suspected tumor recurrence or progress, they were required to return to our hospital for further diagnosis.

Statistical analysis
Categorical variables are shown as the number and percentage of patients. For binary variables, baseline categorical clinical parameters were compared by χ²-test or Fisher's exact test, and numerical values were compared using Student's t-test. Signi cant risk factors for SYN-CRLM and MET-CRLM were analyzed by logistic regression analysis. A log-rank test was used to determine OS and LDFS in each group. The hazard ratio (HR) and the corresponding 95% con dence interval (CI) were estimated with a mixed effects Cox regression model. Statistical analysis was performed by SPSS 22 (IBM company) and GraphPad Prism 8. A two-tailed P-value less than 0.05 was considered a signi cant difference.

Baseline characteristics
From January 1, 2013, to December 31, 2015, 3914 CRC patients were con rmed of having CRLM in our hospital. 782 patients were excluded because of the lack of HBVM results and the inability to evaluate preoperative liver metastasis. A total of 3132 patients were included in nal analysis, which included 1922 men and 1210 females. The median age was 56 years (±11.17, range from 17 to 95). There were 1526 patients with rectal cancer and 1606 patients with colon cancer. According to the HBVM results, patients were divided into group A, B or C. Group A was the HBsAg-positive group, with a total of 413  Table 1 Differences between HBV infection groups The statistical results showed that there was no difference in tumor location, tumor differentiation, T stage, N stage, CEA and CA19-9 in different HBV infection groups. In the male proportion, there was a signi cant difference between group A (62.95%), group B (65.20%) and group C (59.88%) (P = 0.04). The proportion of patients younger than 56 years old in group A (56.90%) was higher than that in group B (43.89%) and group C (45.65%) (P < 0.01). The abnormal proportion of AST (P < 0.01), ALT (P < 0.01) and FIB-4 (P < 0.01) in group A was signi cantly higher than that in group B and group C, with statistical differences. See Table 2 Among the 3132 patients with CRC (before operation), the incidence of SYN-CRLM in group A, group B and group C was 16.95% (70 / 413), 11.60% (74 / 638) and 13.50 (281 / 1800), respectively, with statistical difference(P=0.05). See Table 3. Of the 2707 patients without SYN-CRLM, 92 were lost because of refusing surgery or transferring to other hospitals. 2615 patients underwent primary tumor resection and were followed up. The incidence of MET-CRLM in group A, B and C was 16.51% (54 / 327), 11.53% (64 / 555) and 16.50% (286 / 1733), respectively, with statistical difference(P=0.02). See Table 4.

Risk factors for SYN-CRLM
In this study, a total of 425 people was found to have SYN-CRLM at the rst diagnosis of CRC before surgery. Signi cant differences in tumor location, pathology differentiation, T stage, N stage, HBV infection, CEA, CA19-9, AST, ALT between patients with and without SYN-CRLM were found. There were no signi cant differences in sex, age or FIB-4 among the SYN-CRLM group and the non-SYN-CRLM groups. In multivariate analysis, high or moderate differentiation (P=0.01, HR = 0.62 (0.44-0.87)) was an independent protective factor for SYN-CRLM. The independent risk factors for SYN-CRLM were T stage of In the Cox regression analysis, results showed that better pathological differentiation (P=0.04, HR=0.77 (0.61-0.99)) were independent protective factors. The presence of extrahepatic organ metastasis (P<0.01, HR=10.12 (7.90-12.95)), initial CA19-9 level at the rst diagnosis of CRC (P<0.01, HR=1.50 (1.19-1.89)) and chemotherapy (P<0.01, HR=3.11 (1.91-5.04)) were independent risk factors.
There were also differences in the HBV infection groups (P=0.02). What was different from SYN-CRLM, we found that HBsAg negative infection in group B was an independent protective factor to MET-CRLM (relative to group A, P=0.04, HR= 1.47 (1.02-2.10) and group C, P<0.01, HR=1.49 (1.14-1.95)). See Table 6.

5-year OS and LDFS between different HBV infection groups
In this study, the follow-up period ended on January 1, 2020. A total of 2893 patients were followed up with a median follow-up time of 43 months (range from 0-73). 239 patients were lost. A total of 587 people died, 2306 survived during the follow-up. The 22 . None of the above studies considered that HBV infection might have different effects on CRLM before and after surgery, and these studies did not analyze the effects caused by different HBV infection statuses independently. We believe that is the reason why they have different conclusions.
According to the examination at the rst diagnosis of CRC, SYN-CRLM was found in 425 patients, accounting for 13.57%, 2707 patients did not have SYN-CRLM. The incidence of SYN-CRLM in group B (P<0.01, HR=0.19, 95% CI (0.10-0.33)) and group C (P<0.01, HR=0.28, 95% CI (0.17-0.46)) was signi cantly lower compared to group A. This indicates that the active replication of HBV may increase the risk of SYN-CRLM, and this result is the same as that found by Huo et al. 22 . Abnormal elevation of AST and ALT were independent risk factors for SYN-CRLM, suggesting that SYN-CRLM may be related to HBV-induced hepatocyte damage.
After resection of primary tumor in 2615 patients, MET-CRLM was found in 404 patients (15.45%) during the follow-up. However, different from the effect on SYN-CRLM, HBV infection (including groups A and B) was an independent protective factor relative to group C in the Cox regression analysis. The 5-year LDFS rate of group B was 88.47%, which was signi cantly better than that of group A (83.49%) and group C (83.50%) (P=0.01). This conclusion is similar to that from many previous studies [17][18][19][20] . This result suggested that the liver's immune status may reduce the risk of CRLM in patients who do not have HBV replication or have recovered from HBV infection.
When the cancer cells from the primary sites in the colon escape into the bloodstream, the most likely location where they are lodged is the liver. Kelly et al. suggested that micro-metastasis occurred when cancer cells from the primary CRC escape from the primary location into the portal circulation 27 . The liver immune system removes the tumor cells at the beginning of metastasis 28 . However, in HBsAg-positive patients, the liver immune function was de cient. Peng et al. observed upregulation of the PD-1/PD-L1 pathway in patients with CHB, thereby inhibiting the function and expression of interferon-r (INF-r) and CD8+ T cells, resulting in sustained liver cell damage 29 . Previous studies have suggested that the pathogenesis and metastasis of CRC are related to the upregulation of the PD-1/PD-L1 negative regulatory signaling pathway, which leads to immune escape [30][31][32] . Therefore, we believe that in HBsAg positive CRC patients, the liver immune function is impaired. Before surgery, when the primary tumor continues to enter the liver through blood ow, the consumption of the immune system in the liver is intensi ed, and the incidence of SYN-CRLM is increased. Similar phenomena were observed in primary liver cancer with HBV infection 33,34 .
When the primary tumor is removed surgically, the process of tumor cells continuously entering the liver through the blood is terminated. However, MET-CRLM is still present in about 15% of patients. This is because the liver's immune system is unable to effectively remove residual tumor cells. In this study, the incidence of MET-CRLM in group B was signi cantly lower than that of group A and C. It indicates that there may be some immune response related to HBV infection, which enhances the clearance of tumor cells in the liver.
A number of studies have pointed out that HBV related immune function has been enhanced in patients with HBV infection, and its antiviral function is better than that of patients with chronic hepatitis B infection. Wu et al. noted that PD-1 expression in HBeAg-negative HBV patients was signi cantly lower than that in HBeAg-positive HBV patients 35  The results showed that HBsAg-positive HBV infection increased the risk of SYN-CRLM before surgery, probably by damaging hepatocytes and the liver immune system. HBsAg-negative HBV infection decreased the risk of MET-CRLM after surgery, probably by a good liver immune response. Compared to previous studies, our results can better explain the relationship between HBV and CRLM in the same cohort. However, this study is a single center retrospective analysis, and lack of HBV-DNA data. The occurrence of CRLM may be related to the primary tumor, HBV infection statues, and liver immunology, but the mechanism remain unknown. Therefore, we will carry out a multi-center prospective observational study in the future work, collect complete clinical and experimental data, and further study the mechanism of how HBV infection affects the occurrence of CRLM.