High viral load is a well-known risk factor for HCC, with a dose-response relationship above a level of 2,000 IU/mL, according to the REVEAL-HBV study.15 In contrast, HBsAg seroclearance is associated with favorable clinical outcomes, with reduced risk of HCC especially if age of achieving such endpoint is below 50 years old.16 Treatment with antiviral therapy can effectively suppress HBV DNA and thereby reduce the risk of HCC.17 Of note, nucleoside analogues (NUC) cannot increase the rate of HBsAg seroclearance, which happens in to 1% patients every year.18 A number of novel compounds are being developed to enhance the probability of achieving HBsAg seroclearance, but it remains to be explored whether the benefits from natural HBsAg seroclearance can be extrapolated to novel compound-induced HBsAg seroclearance.

Apart from bringing down the viral load, NUC demonstrates extra effect through reduction of HBV integration to reduce hepatocarcinogenesis.17 In a randomized controlled trial that compared 3 years of tenofovir disoproxil fumarate (TDF) against placebo in a group of patients not meeting treatment criteria, TDF led to 3.28 fold reduction of HBV integration, compared to 1.81 fold reduction in the placebo group (P=0.0037).19,20 With these data, and other evidence supporting the more liberal use of NUC, the threshold to start antiviral therapy has been lowered as reflected by the recently updated World Health Organization guideline that recommends treatment in patients with alanine aminotransferase (ALT) one time above upper limit of normal (ULN) and HBV DNA >2,000 IU/mL, regardless of HBeAg status.21 Similar direction of recommendations is also observed in the recently published Chinese guidelines that suggest antiviral treatment in anyone with detectable HBV DNA with ALT >1 time ULN, or patients with family history of HCC, or aged >30 years old, or presence of significant fibrosis or inflammation.22

Compared to HBV mono-infection, co-infection with human immunodeficiency virus (HIV) accelerates liver disease progression, have higher levels of HBV DNA, and higher risk of cirrhosis and HCC.23 Among HBV-HIV coinfected subjects, HBV DNA >200 IU/mL is associated with hazard ratio (HR) of 2.22, antiretroviral therapy with regimens that are also active against HBV is associated with reduced risk of HCC (HR, 0.42).24 Coinfection with hepatitis C virus and hepatitis delta virus is similarly associated with at least 2-fold increase risk of HCC compared to HBV mono-infection.25,26 It is therefore recommended to routinely screen for these co-infections among patients with CHB and manage accordingly to minimize the risk of disease complications.

Metabolic dysfunction-associated steatotic liver disease (MASLD) is the most common chronic liver disease. At least one-third patients with CHB also suffers from concomitant MASLD.27 The effect of concomitant hepatic steatosis on the risk of HCC in CHB is inconclusive. While some studies reported a protective effect, others report a heightened risk in those with concomitant MASLD and CHB. It is generally believed that MASLD per se is a risk factor for HCC, and concomitant MASLD + CHB should lead to additive if not synergistic risks of HCC. This has been confirmed in biopsy-proven cohorts with MASLD/metabolic dysfunction-associated steatohepatitis.28 However, more recently, an opposite effect has been reported, where imaging-identified hepatic steatosis was observed to be associated with reduced risk of HCC.29,30 The discrepancy is likely related to the population characteristics, in which the modality to diagnose hepatic steatosis contributed to the heterogeneity of clinical course. Study populations that utilized liver biopsy (likely with a clinical indication for biopsy, such as raised ALT or significant fibrosis) to diagnose MASLD had a more unfavorable risk profile than populations that utilized non-invasive means or diagnosis coding to identify hepatic steatosis (more likely incidental finding).31 The effect of metabolic dysfunction is distinct from the effect of hepatic steatosis. The driver for HCC was predominantly metabolic dysfunction, while pure hepatic steatosis was associated with reduced risk of HCC.32 How MASLD interacts with CHB in HCC risk remains to be elucidated.

Type 2 diabetes mellitus (T2DM) is increasingly common among CHB patients. Owing to the aging population in many regions, the prevalence of T2DM has almost doubled from 10.6% to 20.1% in the last 2 decades.33 T2DM is a known risk factor for HCC even in non-HBV infected population,34 and has synergistic effect for HCC in CHB patients.35 The significance of T2DM is highlighted by the incorporation of this factor in the HCC risk prediction scoring system cirrhosis, age, male sex, T2DM (CAMD) which has been externally validated.36 Among those with CHB + T2DM, poor glycemic control as reflected by higher glycated hemoglobin (HbA1c), longer duration of T2DM and lower time reaching target HbA1c is shown to further increase the risk of HCC.37 The drug used for treatment of T2DM has also been implicated in HCC risk among patients with CHB. Compared to non-users, use of sodium glucose co-transporter 2 inhibitor is shown to reduce HCC risk (HR, 0.54; 95% confidence interval, 0.33-0.88) among patients with co-existing T2DM and CHB even after matching for age, gender, metabolic factors, cirrhosis and duration of antiviral therapy.38 Another class of drug commonly used in patients with T2DM or MASLD is statins, a lipid-lowering agent, which is associated with reduced risk of HCC with a dose-dependent manner.39 These observations prompt further evaluation, especially mechanistic studies, to prove a genuine causal relationship.

Cigarette smoking is common among patients with chronic liver disease (to 40%), and negatively impacts incidence and severity of liver fibrosis progression and HCC risk.40 It is associated with at least 2-fold increase in the risk of HCC among people with T2DM + CHB.37 The underlying mechanisms for cigarette smoking in hepatocarcinogenesis are complex, which involve increased pro-inflammatory cytokines, expression of hepatic cancer cells, epithelial to mesenchymal transition and telomere dysfunction.40 Similarly, harmful alcohol use increases the annual incidence of HCC among CHB patients with cirrhosis from 4.1% to 9.9%.41 In contrast, moderate coffee consumption is a protective factor against HCC development in CHB subjects.42,43

To reduce morbidity and mortality from HCC, one of the key secondary prevention measures is early identification of HCC by surveillance. Current guidelines advocate ultrasound scan with or without serum alpha feto-protein (AFP) measurement every 6 months among CHB patients with cirrhosis or risk factors for HCC.5,44 Risk factors for HCC vary between regions, but in general refer to cirrhosis, men ≥40 years old or women ≥50 years old, or family history of HCC. However, AFP is not a perfect tumor-specific marker, as up to 30% HCC are non-AFP secretors.45 Also, AFP is not specific for HCC but could be elevated in other conditions such as hepatitis and germ cell tumor.

A number of alternative tumor-specific markers has been reported to aid early diagnosis of HCC (Fig. 2). Protein-induced by vitamin K absence or antagonist-II (PIVKA-II), also known as des-γ-carboxyprothrombin (DCP), is a prothrombin precursor that would normally be converted to native prothrombin via the action of a carboxylase that is vitamin-K dependent. In malignant hepatocytes, this carboxylase is lacking, resulting in accumulation of PIVKA-II.46 PIVKA-II is much more sensitive than AFP,47 and a combination of both PIVKA-II and AFP is superior to AFP or PIVKA-II alone, with pooled sensitivity and specificity of 82% and 85%, respectively with values of the area under the curve (AUC) for DCP, AFP, DCP + AFP, respectively, being 0.88, 0.75, and 0.90.48 In the Asia-Pacific expert panel consensus in HCC surveillance and monitoring, 88.2% agreed the approach to combine PIVKA-II and AFP for detection of HCC.49 Important considerations should be given among patients who are on vitamin K antagonists or with cholestasis, as these factors might influence PIVKA-II levels.47,50 The glycoform of AFP, the Lens culinaris agglutinin-reactive AFP (AFP-L3), is produced by malignant hepatocytes. AFP-L3 combined with AFP and PIVKA-II is more sensitive and specific to detect early HCC than AFP alone.51 However, criticisms arise for the lack of additional benefit to add AFP-L3 to AFP + PIVKA-II in the scoring systems developed to predict HCC (gender, age, AFP-L3, AFP, DCP [GALAD] and gender, age, AFP, DCP [GAAD] had similar performance characteristics).52 Nevertheless, a substantial proportion of HCC express AFP-L3 alone but not AFP nor PIVKA-II.53

Tumor-specific protein 70 (SP70) is an antigen with relative molecular mass of 70 kDa. It is a tumor-derived protein that is involved in gene regulation for promoting cancer cell proliferation and metastasis. SP70 was initially found to be raised in patients with non-small cell lung cancer, and thus is not specific for HCC.54 In patients with HCC, combination of SP70 with AFP improved the diagnostic accuracy of HCC (AUC, 0.909). Elevated SP70 is also associated with worse recurrence-free survival after curative treatment for HCC.55 Like many other tumors, HCC is associated with DNA methylation abnormalities. Certain methylation changes are involved in carcinogenesis, such as promotor-region methylation in CpG islands. With advancement in technology, the methylated DNA markers could be enriched and detected in plasma isolated from cell-free DNA.56 A recent study identified a multi-target HCC panel consisting of AFP, AFP-L3, 3-methylated DNA markers and B3GALT6 (a galactosyltransferase) as a novel blood-based biomarker panel to detect early HCC.57 Tumor cells are known to release nanometer-sized extracellular vesicles (EV) that are membrane-derived and function by transferring donor cell-derived bioactive molecules. It is recently reported that circulating EVs enriched with polymeric immunoglobulin receptor, known to be aberrantly expressed in cancer tissue, were elevated in plasma of patients with late-stage HCC and contributed to cancer stemness and tumorigenesis.58

Apart from PIVKA-II and AFP-L3 with commercialized assays available, other tumor-specific markers are currently used as research tool only owing to limited accessibility, need of sophisticated sample processing and lack of procedure/assay standardization.

In cirrhotic livers, imaging diagnosis of HCC can be achieved by typical arterial hyperenhancement and portovenous washout among lesions >10 mm. Such criteria has been endorsed by European Association for the Study of the Liver (EASL) clinical practice guidelines5 and forms the backbone of the liver imaging reporting and data system (LI-RADS) endorsed by American Association for the Study of Liver Diseases (AASLD) guidelines to stratify the probability of a lesion being an HCC, where LR-5 refers to definite HCC.59 The sensitivity and specificity for HCC was found to be similarly at to 60% and to 90%, respectively, for imaging-based diagnosis using the EASL criteria or LR-5.60 The role of artificial intelligence in enhancing the accuracy of radiological diagnosis and characterization of HCC has been explored, such as the use of convoluted neural network to assist in assigning LI-RADS grades,61 histological grades,62 or radiomics approach via quantifying radiological characteristics with machine learning models such as texture analysis and topological data.63 These approaches are considered exploratory at this juncture pending extensive evaluation and validation. The bottom line is the careful use of contrast-enhanced cross-sectional imaging for sizable lesions in the background of cirrhotic liver. Histological diagnosis is still necessary for lesions arising from non-cirrhotic livers, indeterminate radiological features, and for the practice of precision medicine.64

To identify high-risk individuals for HCC even before its onset, i.e., HCC prediction, many scores have been developed for HCC risk stratification in patients with CHB. These scores comprise of mainly clinical parameters. For instance, sex, age, serum levels of ALT, HBeAg status, HBV DNA level (REACH-B),65 age, gender, core promotor mutation, HBV DNA level, and cirrhosis (GAG-HCC),8 age, albumin, bilirubin, HBV DNA level, cirrhosis (CU-HCC),10 liver stiffness, age, serum albumin, HBV DNA level (LSM-HCC),66 CAMD score (discussed above),36 age, gender, platelet count (PAGE-B),67 and modified REACH-B, liver stiffness instead of HBV DNA level.68 Age, gender and platelet are commonly incorporated in these scores (Fig. 2). Table 1 summarizes the risk scores reported so far for prediction of HCC in CHB patients. Although with acceptable discriminating power,69 none of these scores are currently recommended by clinical guidelines for routine use for a few reasons. Firstly, not all scores have been externally validated and since the majority of scores were developed in Asian patients, the performance among Caucasian patients is suboptimal, let alone some ethnicities such as African patients (Table 1). Secondly, most scores reported a high negative predictive value of 95-99% but a low positive predictive value from 10% to 40% (Table 1). Thirdly, certain parameters are prone to subjective interpretation (e.g., presence of cirrhosis on ultrasound) or suffer from limited availability (e.g., HBV mutations, HBV genotypes), with complicated mathematical formula to calculate a final score. Lastly, the implications of dynamic change in predictive scores with time and treatment on the risk of HCC have been unclear.

Apart from these scores that focus on the clinical parameters, other variables predictive of HCC might also play a role. These include viral biomarkers and liver fibrosis markers (Fig. 2).