Hepatitis B virus infection and active replication promote the formation of vascular invasion in hepatocellular carcinoma

Background Vascular invasion, including microvascular invasion (MVI) and portal vein tumor thrombus (PVTT), is associated with the postoperative recurrence of hepatocellular carcinoma (HCC). We aimed to investigate the potential impact of hepatitis B virus (HBV) activity on the development of vascular invasion. Methods Patients with HBV and tumor-related factors of HCC who had undergone hepatectomy were retrospectively enrolled and analyzed to identify the risk factors for developing vascular invasion. Results A total of 486 patients were included in this study. The overall proportion of patients with vascular invasion, including MVI and PVTT, was 60.3% (293/486). The incidence of MVI was 58.2% (283/486) whereas PVTT was 22.2% (108/486). Univariate analysis revealed that positive Hepatitis B virus surface Antigen (HBsAg) was significantly associated with the presence of vascular invasion. In a multivariate regression analysis carried out in patients with HBV-related HCC, positive Hepatitis B virus e Antigen (HBeAg)(OR = 1.83, P = 0.019) and a detectable seral HBV DNA load (OR = 1.68, P = 0.027) were independent risk factors of vascular invasion. The patients in the severe MVI group had a significantly higher rate of positive seral HBsAg (P = 0.005), positive seral HBeAg (P = 0.016), a detectable seral HBV DNA load (> 50 IU/ml) (P < 0.001) and a lower rate of anti-viral treatment (P = 0.002) compared with those in the mild MVI group and MVI-negative group. Whereas, HCC with PVTT invading the main trunk showed a significantly higher rate of positive HBsAg (P = 0.007), positive HBeAg (P = 0.04), cirrhosis (P = 0.005) and a lower rate of receiving antiviral treatment (P = 0.009) compared with patients with no PVTT or PVTT invading the ipsilateral portal vein. Patients with vascular invasion also had a significantly higher level of seral HBV DNA load than patients without vascular invasion (P = 0.008). Conclusions In HCC patients, HBV infection and active HBV replication were associated with the development of vascular invasion.


Background
Hepatocellular carcinoma (HCC) is the fifth most common cancer and the third leading cause of cancer-related death in the world [1]. Although surgical resection and liver transplantation could offer a promising prognosis for selected patients with HCC, the high postoperative recurrence rate has impaired long-time survival. Among various risk factors, vascular invasion, including microvascular and macrovascular invasion, has been proven to be an independent factor predicting high recurrence and poor survival rate [2][3][4]. Microvascular invasion (MVI) was defined by most studies as microscopically confirmed tumor cell clusters within a vascular cavity lined with endothelium adjacent to the tumor [5,6]. Conversely, macrovascular invasion mostly occurs in the portal vein system and is known as a portal vein tumor thrombus (PVTT); a PVTT can be identified during imaging examination or intraoperative exploration. A large tumor size, multinodular lesion, elevated level of desc-carboxy prothrombin (DCP) and certain imaging characteristics were reported to be factors predicting the presence of MVI, whereas the tumor size, Edmondson-Steiner histological grading, number of nodules and α-fetoprotein (AFP) level were associated with PVTT [2,5,7,8].
Chronic hepatitis B virus (HBV) infection is a major risk factor for the development of liver cirrhosis and HCC, especially in East Asia [6]. HBV-related factors, such as seropositivity of hepatitis B e-antigen (HBeAg), high hepatitis B surface antigen (HBsAg) level and high serum HBV DNA load, were found to be significantly related to an increased risk of HBV-associated cirrhosis and HCC [9,10]. These factors were also reported to be associated with an increased recurrence rate and a decreased survival rate of HCC after surgical resection [11,12]. Fundamental research has revealed that the HBV-initiated tumorigenic process may play a role in the development of the vascular invasion of HCC [13][14][15]. Recently, Lei Z et al. established a nomogram for preoperative prediction of the presence of MVI in HBV-related HCC, in which a high HBV DNA load (>10 4 IU/ml) was independently associated with the development of MVI [16]. These findings indicated a potential correlation between active HBV replication and the development of vascular invasion in HCC. To the best of our knowledge, no published study has provided insight into this issue. Therefore, we conducted a clinical study to further explore the impact of HBV-related factors on the formation of vascular invasion in HCC.

Study population
This was a retrospective study based on a prospectively compiled clinical and pathology database at a treatment center for HCC with PVTT at the Eastern Hepatobiliary Surgery Hospital, Shanghai, China. The study was approved by our Institutional Review Board, and written informed consent was obtained from all patients for their data to be used in this research.
HCC patients who had undergone surgical resection and confirmed by pathological examination at our center were included in this study. Exclusion criteria included hepatitis C virus (HCV)-related HCC, preoperative transarterial chemoembolization (TACE) or radiotherapy, non-curative resection, recurrent lesions, and a lack of complete clinical or pathological information.
For patients included in the study, the following clinical data and pathological results were collected: (1) demographic data, including age and gender and history of antiviral treatment; (2) results of preoperative laboratory blood tests, including HBsAg, HBeAg, HBV-DNA level, AFP, DCP, albumin, total bilirubin, alanine aminotransferase, and aspartate aminotransferase; and (3) imaging and pathologic findings, including the presence and classification of PVTT, maximal tumor size, tumor number, capsule, presence and classification of MVI, and presence of cirrhosis.
Tests for the viral replication status, including those for HBsAg and its antibody, HBeAg and its antibody, and HBcAb, were performed. The serum HBV-DNA level was quantified by the polymerase chain reaction assay (ABI 7500; Applied Biosystems, Foster City, CA, USA) with a linear range of quantification of 50 to 2,000,000 IU/ml. The lower limit of detection was 50 IU/ml. Patients who had received standard interferon therapy or had been using oral anti-viral drugs for a duration of more than 2 months before surgery were classified as the anti-viral treatment group.

Diagnostic criteria of vascular invasion
The diagnostic criterion of MVI was the presence of a tumor cell nest in the portal vein, hepatic artery, hepatic vein, bile duct or lymph duct in the tumor surrounding the liver tissue under microscopic examination [2,17]. The number and distribution of invaded vessels were measured to divide the patients with MVI into two groups as follows: patients in the mild MVI group (M1) had 1 to 5 involved vessels distributed within a 1-cm area from the tumor margin, whereas patients in the severe MVI group (M2) had more than 5 vessels invaded or had invaded vessels located more than 1 cm from the tumor margin. Every specimen was reviewed independently by two senior hepatobiliary pathologists to detect MVI. If the two pathologists had an inconsistent diagnosis, the findings were discussed to reach a final decision.
All HCC patients admitted to our center underwent a routine three-phrase dynamic CT or MRI examination before any treatment was carried out. PVTT was diagnosed when there were low-attenuation intraluminal masses that expanded the portal vein, or filling defects in the portal vein system, as presented in CT or MRI imaging. PVTT was confirmed and reassessed by palpation or ultrasound during operation. The final diagnosis was dependent on the intraoperative or pathologic findings. PVTT was classified according to Cheng's classification, which has been shown to be effective in stratifying the severity of PVTT as follows: type I, invasion of the tumor thrombus into the segmental or sectoral branches of the portal vein or above; type II, involvement of the right or left portal vein; type III, invasion of the main trunk of the portal vein; and type IV, involvement of the superior mesenteric vein.

Statistical analysis
All calculations were performed using Stata 12.0 software (StataCorp, Texas 77,845, USA). Continuous and categorized data were compared using Pearson's chi-squared test, Fisher's exact test, or Student's t test, as appropriate. Binary logistic regression was used to evaluate the relationship between the presence of vascular invasion as the dependent variable and factors that were significant in the univariate analysis as independent variables, using the stepwise backward method (Wald). The enter limit and remove limit were P = 0.05 and P = 0.10, respectively. Because viral factors, including seral HbeAg, the seral HBV DNA load, presence of cirrhosis and usage of antiviral treatment, were only meaningful when patients had HBV infection, only patients with positive seral HbsAg were included in multivariate analysis. A P < 0.05 was considered to indicate statistical significance.

Results
From May 1, 2015 to July 31, 2016, 675 patients with a preoperative diagnosis of HCC who underwent surgical resection at our center were identified. After careful examination, 189 patients were excluded, including 77 for preoperative TACE or radiotherapy, 36 for being diagnosed with a histological type other than HCC, 7 for HCV infection, 27 for recurrent lesions, 21 for non-curative resection, and 21 for failure to obtain detailed clinic information. Finally, 486 HCC patients, 422 men and 64 women, with a median age of 52 years (range, 22-80 years), fulfilled the inclusion criteria and were enrolled in the study. Most

Univariate analysis of viral and tumor factors predicting vascular invasion in HCC
Univariate analysis revealed that virus-related serum markers, including positive HBsAg (P = 0.005), positive HBeAg (P < 0.001) and a detectable HBV DNA load (P < 0.001), were significantly associated with the presence of vascular invasion, whereas vascular invasion was less frequently detected in patients in the anti-viral drug group (P = 0.003). The significant viral factors predicting MVI were the same as those of vascular invasion for patients with vascular invasion and MVI and were mostly overlapping. Similarly, virus-related seral markers, including positive HBsAg (P = 0.003), positive HBeAg (P = 0.005), a detectable HBV DNA load (P = 0.025), and the presence of cirrhosis (P < 0.001), were significantly associated with the presence of PVTT, whereas patients who were undergoing anti-viral treatment (P = 0.015) had a significantly lower risk of developing PVTT (Table 1).

Multivariate analysis of viral and tumor factors predicting vascular invasion in HCC
Multivariate logistic regression analysis was carried out in patients with positive seral HbsAg, utilizing binary variables that were significant in the univariate analysis. As shown in Table 2, positive seral HBeAg (OR = 1.83, P = 0.019) and a detectable seral HBV DNA load (OR = 1.68, P = 0.027) were independent risk factors of vascular invasion in the multivariate regression analysis. Moreover, tumor-related factors, including a seral AFP level > 20 ng/ml (OR = 2.51, P < 0.001), multiple lesions (OR = 2.18, P = 0.038), tumor size >3 cm (OR = 1.73, P = 0.035), Edmonson grades III/IV (OR = 2.48, P = 0.013) and incomplete/absent tumor capsule (OR = 2.17, P = 0.006), were significantly and independently associated with vascular invasion. Factors predictive of MVI were similar to those predictive of vascular invasion, except that the impact of seral HBeAg on the formation of MVI didn't reach statistical significance (OR = 1.59, P = 0.059). Regarding the risk factors of PVTT, the impact of the seropositivity of HBeAg (OR = 1.67, P = 0.046), tumor diameter > 3 cm (OR = 8.86, P < 0.001), incomplete or absent encapsulation (OR = 3.59, P = 0.003) and DCP > 100 mAU/ml (OR = 2.90, P = 0.022) were significant in the multivariable analysis. Table 3 shows that the severe MVI group had a significantly higher rate of positive seral HBsAg, positive seral HBeAg, a detectable seral HBV DNA load (> 50 IU/ml), as well as a lower rate of antiviral treatment, compared with the mild MVI group and negative group. By contrast, for the classification of PVTT, HCC with type III/IV PVTT had a significantly higher rate of positive seral HBsAg, positive seral HBeAg, and cirrhosis, as well as a lower rate of receiving antiviral treatment compared with the type I/II group and PVTT-negative group. Patients with vascular invasion had a significantly higher seral HBV DNA load than patients without vascular invasion (Table 4).

Discussion
The presence of vascular invasion, including MVI and PVTT, was significantly associated with a high risk of postoperative recurrence, which is a major obstacle to improving the prognosis of HCC [6,17,18]. However, the risk factors and underlying mechanism leading to the formation of vascular invasion remain largely unknown. In East Asia, the majority of HCC develops within an environment of chronic inflammation caused by HBV infection. Recently, the results of fundamental studies have indicated that the HBV status is a potent etiological factor predisposing HCC patients to develop vascular invasion. HBV X protein (HBx), a key regulatory multifunctional protein of the virus, has been reported to be involved in the development of MVI and is associated with postoperative recurrence [14,19,20]. Yang et al. found that the seropositivity of HBsAg was associated with a high risk of developing PVTT, and the activity of the TGF-β-miR-34a-CCL22 axis induced by the change in the liver microenvironment caused by HBV infection may play an important role in the   [15]. The potential correlation between HBV replication and the formation of MVI in HCC have also been studied in some preliminary clinical studies. Chen et al. retrospectively studied the impact of ascites, as well as tumor-and HBV-related factors, on the formation of vascular invasion and found negative results concerning the impact of viral factors; however, it is worth noting that the limited number of cases with MVI (n = 12) and incomplete data concerning the status of HBV infection may limit the power of their results [21]. To establish a preoperative prediction model for MVI, a large cohort of HBV-related HCC patients (n = 1004) was analyzed by Lei et al., revealing that a high seral HBV DNA load (> 10 4 IU/ml) was an independent factor predicting the presence of MVI. The other predictive variables were well-known tumorrelated factors, including a large tumor diameter, multiple nodules, an incomplete capsule and an AFP level > 20 ng/ml [16]. Nevertheless, the relationship between HBV infection and vascular invasion has rarely been intentionally researched in a well-designed clinical study.
Our study was based on a prospectively collected database with comprehensive data indicating the status of HBV infection and vascular invasion. The results showed that compared with patients without HBV infection, the incidence of vascular invasion, including MVI and PVTT, was significantly increased in HBV-related HCC. In the multivariate analysis carried out in positive HBsAg patients, positive HBeAg and a detectable seral HBV DNA load (> 50 IU/ml) were significantly associated with development of vascular invasion. In addition, our results revealed that in HBV-related HCC patients, a more severe level of vascular invasion was associated with a higher rate of active HBV replication, as reflected in positive HBeAg or a detectable HBV DNA load. These findings provided promising clinical evidence to demonstrate that in addition to tumor-related factors, the activity of HBV infection plays a key role in the development of vascular invasion in HCC patients.
The postoperative recurrence of HBV-related HCC was categorized into two groups, early and late recurrence, with a cut-off time at 2 years [22]. Late recurrence (> 2 years after resection) usually presented as a metachronous tumor with different genetic and histological features from the primary HCC [22,23]. It was revealed that HBV-related factors, including a high hepatic inflammatory activity score and high HBV DNA load, were significantly associated with late recurrence, whereas sustained suppression of HBV replication by anti-viral drugs achieved a lower rate of late recurrence [12,24,25]. Tumors occurring within 2 years after surgery were classified as early recurrence, which was strongly associated with tumor-related factors, including tumor size and the presence of nodules, vascular invasion and resection margin [22]. A randomized controlled trial by Lin et al. revealed that patients receiving anti-viral treatment showed a significantly better 2-year overall (93.8% vs 62.2%) and recurrence-free (55.6% vs 19.5%) survival [26]. However, it is difficult to understand the effect of anti-viral drugs on inhibiting early postoperative recurrence (< 2 years after resection), which was considered the result of the regrowth of micro-metastases in the liver that were not detected and resected during the operation. Our result demonstrates that active HBV replication is associated with a high rate of vascular invasion in HCC patients, which may partially explain the anti-tumor effect of antiviral treatment. We could speculate that the suppression of HBV replication via antiviral treatment might decrease the invasiveness and metastatic potential of HCC to reduce the risk of early postoperative recurrence. The seral HBV DNA load is usually divided into high and low levels at a cut-off value of 2000 IU/ml. In this study, the impact of the seral HBV DNA load on the formation of vascular invasion was not significant if the cut-off value was set at this point. This result implies that the correlation between the HBV DNA load and occurrence of vascular invasion is not linear. Additionally, this result may also be caused by a proportion of patients with a high HBV DNA load being in the "immune tolerant" phase with no or mild substantial liver injury [10]. According to the current guidelines, anti-viral drugs should be prescribed in chronic hepatitis B patients with a serum HBV DNA load above 2000 IU/ml and elevated ALT levels, in the absence of sufficient evidence of cirrhosis [10]. However, for patients with a low HBV DNA level (< 2000 IU/ml) without advanced liver disease, the benefit of anti-viral treatment has not been well clarified. In this study, HCC patients with an undetectable HBV DNA load (≤ 50 IU/ml) had a lower incidence of vascular invasion than patients with a detectable HBV DNA load (> 50 IU/ ml). These results suggested that it may also be beneficial to receive anti-viral drugs for patients who do not meet the current treatment indication.
The suppressed HBV replication by anti-viral treatment was supposed to correlate with a lower rate of vascular invasion. However, the inhibitory effect of antiviral treatment on the development of vascular invasion was overshadowed by tumor-related factors in the multivariate analysis. The following reasons may explain this phenomenon. First, patients who have undergone antiviral treatment usually have no or mild cirrhosis with normal liver function [27,28]. Surgeons are more likely  to apply hepatectomy with a wider resection margin to these patients. Theoretically, a wide surgical margin will lead to a higher detection rate of MVI during pathological evaluation. Second, patients who have undergone anti-viral treatment have enjoyed good health care and regular surveillance, leading to early detection of HCC.
Thus, the inhibitory effect of anti-viral treatment tended to be overshadowed by the early tumor features. It is worth noting that MVI and PVTT have different risk factor profiles in our research. For HBV-related factors, only detectable HBV load was associated MVI, while only positive HBeAg was associated with PVTT. Although MVI and PVTT were two common types of vascular invasion of HCC, there was no evidence indicating potential causal relationship between them. Previous clinic studies also revealed that MVI and PVTT had inconsistent predicting factors [2,5,7,8]. Further fundamental and clinic studies are needed to clarify the relationship between MVI and PVTT. This study might not be able to reveal the full landscape of the relationship between HBV activity and the occurrence of vascular invasion in HCC. In particular, because of the limited follow-up time, we failed to carry out a survival analysis to determine significant factors contributing to recurrence or survival. However, the main finding of this study is the association between HBV infection status and presence of vascular invasion in HCC, lack of survival information may have a less impact on our conclusion. Additionally, inconsistency existed between the protocol of anti-viral treatment and surgical procedures because of the retrospective nature of this research. Furthermore, the surgical margin varied in patients with different levels of cirrhosis, a finding that might affect the detection rate of MVI. At last, only HBV-related HCC was studied in this research, the conclusion isn't applicable for HCC caused by other hepatic virus. Despite these limitations, we first found the interesting phenomenon that HBV infection and replication status were independently associated with the formation of vascular invasion in HCC, which may partially explain the inhibitory effect of anti-viral treatment on early HCC recurrence.

Conclusions
In addition to characteristics of the tumor itself, HBV infection and active replication were independently associated with the development of vascular invasion in HCC. In patients with HBV-related HCC with positive HBeAg or a detectable HBV DNA load, an increased risk of vascular invasion should be recognized.