3.1.1 HBsAg in the development of the primary HCC lesion

Hepatitis B virus-associated carcinogenesis is a multifactorial process involving direct effects of viral proteins, indirect mechanisms through chronic inflammation and immune evasion, and the integration of HBV DNA.18 Although patients with HBsAg loss may still develop HCC and require surveillance,17 the risk is low and comparable in those who clear HBsAg spontaneously or following NA therapy,19 including in patients with HBsAg loss after finite NA therapy.20, 21 In young patients and those with HBsAg loss prior to the onset of cirrhosis the risk of HCC is minimal.3 Thus, the focus here is on patients remaining HBsAg positive.

Although the majority of HBV-related HCC develops in cirrhotic liver, unlike hepatitis C infection, some cases occur relatively early in the disease course, prior to development of cirrhosis or decompensated disease. This observation in individuals considered to have minimal liver damage is compatible with a direct oncogenic effect of HBV22 and has been cited as a reason for starting antiviral therapy in ‘immune tolerant’ patients,23 despite guidance to the contrary.3 The direct oncogenic effect of HBV may occur via multiple mechanisms. Integration of HBV DNA into the host genome may result in genomic instability, which together with production of HBV proteins, especially the regulatory protein HBx, may result in uncontrolled cell proliferation. Although the role of viral proteins remains unclear,22 Dane particles, HBsAg-expressing SVPs and free HBsAg may contribute directly to tumour development24-26 (Figure 2), potentially via activation of the nuclear factor-ΚB pathway.27

Potential role of HBsAg in development of HCC. Integration of HBV DNA into the host genome may result in genomic instability, which together with production of HBV proteins, may result in uncontrolled cell proliferation and tumour development. LHB, large HBV surface protein; MHB, medium HBV surface protein; SHB, small HBV surface protein; HBcAg, hepatitis B core antigen; HBeAg, hepatitis B ‘e’ antigen; cccDNA, covalently closed circular DNA

Hepatitis B surface antigen consists of large (LHB; pre-S1), medium (MHB; pre-S2), and small (SHB; S) HBV surface proteins, and mutations in the coding regions for these proteins have been shown to increase as HBV-related disease progresses and to confer a high risk of HCC,28 leading to loss of genome integrity,29 with some found to be more common in patients with cirrhosis/HCC compared with those without.30 Recently, SHB has been shown to induce the epithelial-mesenchymal transition process in HCC cells, significantly increasing their migratory and invasive ability as well as their metastatic potential.31 In the same study, in HCC sections from patients who underwent curative resection, expression of SHB in HCC tumours was positively correlated with a more aggressive phenotype, namely higher tumour number, loss of tumour encapsulation and a higher TNM stage, and reduced post-resection survival.31 Mutations resulting in impaired HBsAg secretion and cellular proliferation,29 as well as calcium homeostasis and chromosome instability,32 and altered host gene expression,33 have been identified. As well as potentially having a direct role in carcinogenesis, these mutant HBsAg proteins may have a role to play in the ability of HBV to evade immune detection.

The direct oncogenic potential of HBsAg is supported by preclinical studies and suggests that unmutated HBsAg also has oncogenic potential. In an HBsAg transgenic mouse model, HBsAg accumulation in hepatocytes activated cancer stem cell markers to potentiate HCC development.22 Pre-S1 has been shown to act as an oncoprotein and to play a key role in the appearance and self-renewal of cancer stem cells during HCC development.26 Knockout of HBsAg in an HCC cell line resulted in decreased expression of HBsAg and significant attenuation of HCC proliferation in vitro as well as tumorigenicity in vivo.34 In the same study, overexpression of all three HBsAg surface proteins promoted proliferation and tumour formation.34 HBsAg levels may also lead to endoplasmic reticulum stress and a pre-malignant phenotype.35 Together, these observations suggest that neutralising the effect of HBsAg could have a beneficial effect in HBV-associated HCC.

As well as direct promotion of carcinogenesis, it is also thought that evasion from self-immunity may be necessary for cancer growth. Given the observation that HBsAg carriers have a 25-37-fold increased risk of developing HCC compared with non-infected people, HBsAg may have a function in immune evasion rather than being purely oncogenic,18 and may activate signalling pathways that link inflammation and tumour progression during chronic HBV infection.36 Although HBsAg has a tumour-promoting effect, clearance of HBsAg reduces, but does not eliminate, the risk of HCC entirely,16, 17 and the risk persists particularly in cirrhotic patients, despite HBsAg clearance.

In the Taiwanese REVEAL cohort, relative risk of HCC was shown to be dependent on serum levels of HBsAg.37 Compared with levels <100 IU/mL, patients with levels of 100-999 and ≥1000 IU/mL were found to be 2.8- and 4.1-fold more likely to develop HCC, respectively,37 and in a further study,12 levels of HBsAg ≥ 1000 IU/mL conferred a 13.7-fold increased HCC risk compared with lower levels. This HBsAg level-dependent association was particularly apparent in patients with low HBV DNA,12, 37 indicating the importance of lowering HBsAg even in patients with low HBV DNA. Risk of HCC is also dependent on serum levels of HBV DNA, with cumulative incidence rates of HCC increasing from 1.3% to 14.9% with HBV DNA levels of ≤300 to ≥1 million copies/mL, respectively.38 HBsAg- and HBV DNA-related increases were independent of HBeAg, serum ALT levels or cirrhosis. As a consequence, three cut-offs for HBsAg (<100, 100-999, ≥1000 IU/mL) and three for HBV DNA (<104, 104-106, ≥106 copies/mL) have been included in risk scores for HBV-related HCC.37 It remains unclear, whether the risk of HCC is higher because of increased liver inflammation in response to viral replication or to direct mechanisms. In a further recent analysis of data from the REVEAL cohort in patients remaining HBeAg-positive throughout follow-up, risk of HCC was found to be 3-fold higher among those with declining HBsAg levels (low HBsAg levels at baseline or declining to <10 000 IU/mL) than in those with HBsAg levels persistently above 10 000 IU/mL without decline (P = .03).39

Although the increase in HCC is independent of cirrhosis, the majority of HBV-related HCC is on the background of a cirrhotic liver. Indeed, cirrhosis is the most important risk factor, with the 5-year cumulative risk of developing HCC being 0.6%-2.4% in individuals with untreated chronic hepatitis B compared with 9.7%-15.5% in those with compensated cirrhosis.40 The risk increases further with hepatic decompensation, with 3-year rates of HBV-related HCC in patients with decompensated disease being reported at 25%.41 Therefore, it is the combination of cirrhosis, in particular decompensated cirrhosis, and the serum levels of HBV DNA that particularly predispose to the development of HCC.

Coinfection with the hepatitis delta virus (HDV) is a key risk factor for the development of HCC. Approximately 15-20 million people are infected with HDV, a subviral satellite that requires HBV for propagation.42 In coinfected patients, liver disease is more rapidly progressive, more likely to progress to HCC, and is associated with higher mortality than in those patients infected with only HBV.42, 43 The additional oncogenic capacity of HDV is not fully understood, although there is some evidence that small HDV antigens may have a direct role in carcinogenesis.43 In addition, HDV replication in hepatocyte nuclei results in dysregulation of pathways, increased oxidative stress and epigenetic changes that can lead to malignant transformation.42 Activation and coregulation of genes critically associated with DNA replication, damage, and repair are higher in HDV/HBV-HCC hepatocytes than either non-malignant or HBV-HCC hepatocytes, suggesting that genetic instability may be an important mechanism in HDV/HBV-related HCC and that, despite the dependence of HDV on HBV, distinct molecular mechanisms are involved.44

3.1.2 Recurrence of HCC post-LT

In early-stage HCC, LT is a preferred treatment option.1 Although the prevalence of HBV-related LT in Europe has stabilised over the last 30 years, HBV-related HCC as an indication for LT has increased in contrast to HBV-related decompensated liver disease.45 In addition, the incidence of patients undergoing LT with positive HBV DNA is increasing, and in patients with HBV-related HCC as the indication for transplant, this is a predictor of reduced survival.45 In Asia, with the exception of Japan, HBV-related HCC remains the most frequent indication for liver resection or LT. Although there is debate around whether the likelihood of HCC recurrence is higher in patients with HBV-related disease compared with that of other aetiologies, there is little debate that HCC will recur in a proportion of patients.

Most patients will be HBsAg positive at the time of LT, and rates of HCC recurrence post-LT in these patients have been reported to be around 10%-15%.46 A proportion of patients with HCC recurrence will also be re-infected with HBV. The rate of reinfection is higher than for other HBV-related transplant indications,47 and both HBV reinfection and HCC recurrence are associated with poorer survival.46

It is deemed critical to understand the cause of recurrence of HCC post-LT. There are two hypotheses that can be put forward: firstly, that post-LT residual HCC tumour cell populations, containing integrated HBV DNA, expand and independently replicate HBV, leading to the recurrence of both HCC and HBV47; secondly, that residual HBV RNA or HBsAg produced by other non-tumour cells drive the reactivation of HBV, in turn driving the de novo recurrence of HCC. Data from a recent study in Korea (unpublished data) indicate that the first scenario may be more likely (Figure 3), since HBV DNA positivity was not linked to HBV reactivation, whereas recurrence of HCC was an independent risk factor for HBV reinfection. These data are further supported by observations that patients who become HBV DNA and HBsAg negative after removal of the primary tumour can have a recurrence of HCC expressing HBV DNA,47 indicating that the recurrence of tumour cells alone could drive the recurrence of HBV. In addition, low-level oncogenic HBV variants persist post-LT despite potent anti-HBV prophylaxis and could lead to de novo carcinogenesis, as described earlier.

Potential cause of HCC recurrence and HBV reinfection post-LT. Post-LT residual HCC tumour cell populations containing integrated HBV DNA may expand and independently replicate HBV, leading to the recurrence of both HCC and HBV. LT, liver transplantation; HCC, hepatocellular carcinoma; HBV, hepatitis B virus

The immunosuppression required to minimize graft rejection in patients undergoing LT is also likely to contribute to HCC recurrence post-LT. Tumour progression is more rapid and aggressive in immunosuppressed patients post-LT,48 with the degree of immunosuppression having a negative impact on survival. Corticosteroids used in post-LT immunosuppression protocols stimulate viral replication through binding to the glucocorticoid responsive enhancer region of the HBV genome. As a consequence, it has been proposed that these drugs should be removed from immunosuppressive regimens as soon as possible to reduce the likelihood of HBV reinfection and HCC recurrence.49

3.1.3 Recurrence of HCC post-liver resection

Resection is also a preferred treatment option in patients with early-stage HCC,1 and in Asia, although LT provides the best curative option, a relatively small donor pool means that resection is often the first-line approach. Compared with rates of HCC recurrence post-LT, recurrence rates post-liver resection are much higher, at 56%-70%.50, 51 Whereas it seems likely that the cause of HBV-related HCC recurrence post-LT is expansion of residual tumour cells containing integrated HBV DNA, after liver resection a different disease course is potentially more probable. After resection, cirrhotic liver, which is precancerous per se, will remain, with the potential to undergo malignant transformation regardless of the existence of HBV-derived oncogenic drivers. The presence of residual HBV will further increase the likelihood of HCC recurrence, as with primary tumour formation (Figure 4). This risk may be largely that of late HCC recurrence52, 53 rather than early recurrence,54 with specific mutant antigens conferring a particularly high risk. The presence of pre-S mutants in patients with HBV-related HCC has been associated with a significantly higher risk of HCC recurrence after curative surgical resection.55 Of note, evaluation of predictive factors for HCC recurrence post curative resection for HBV-related early-stage HCC indicated that whereas preoperative HBV DNA levels ≥20 000 IU/mL were predictive of recurrence within 2 years (hazard ratio [HR] 2.77; P < .001), patients were at risk of late recurrence if preoperative HBsAg levels were ≥4000 IU/mL (HR 2.80; P = .023).56 Similar results were reported in patients receiving NAs after surgery; an HBsAg level of >200 IU/mL was associated with a 2-fold increased likelihood of late recurrence (P = .027).57 Most recently, it has been shown that while an HBsAg level >200 IU/mL is an independent predictor for late recurrence (2-5 years post-resection), HBsAg levels >50 IU/mL could predict recurrence and mortality beyond 5 years. This suggests that different cut-off values of HBsAg may be useful to predict outcomes at different intervals after surgical resection.58

HCC recurrence after liver resection. After resection, cirrhotic liver retains the potential to undergo malignant transformation. This is possible in the absence of HBV-derived oncogenic drivers but is increased in their presence. cccDNA, covalently closed circular DNA; HCC, hepatocellular carcinoma; HBV, hepatitis B virus; LHB, large HBV surface protein; MHB, medium HBV surface protein; SHB, small HBV surface protein; HBcAg, hepatitis B core antigen; HBeAg, hepatitis B ‘e’ antigen

Furthermore, in a prospective study of 89 patients undergoing liver resection for HBV-related HCC, circulating HBsAg levels correlated with cccDNA copy number in both tumour and non-neoplastic liver tissue but not with serum HBV DNA, total intrahepatic HBV DNA, viral replicative activity, or transcriptional activity. Correlation of HBsAg with cccDNA copy number, regardless of malignant status, suggests that a larger pool of cccDNA is associated with a higher rate of HBsAg production.59 It is important to distinguish the risk attributable to HBsAg positivity and HBV viraemia at the time of surgical intervention. Although risk of recurrence is high with HBV DNA levels ≥ 104 copies/mL,60 high HBsAg levels despite low HBV DNA levels represent ongoing inflammation and thus are also associated with an increased risk of HCC recurrence.61 It may be expected that in patients with a high viral load, treatment with NAs will decrease HBV DNA levels, reduce liver inflammation resulting from HBV replication and therefore decrease the risk of HCC recurrence. This has been suggested in patients regardless of viral load62, 63 but remains to be proved unequivocally.

A large proportion of patients with HBV-related HCC present late in their disease course, when potentially curative therapies are not an option, and there are currently no data on the association between HBsAg and advanced stages of HCC.