Prediction of Hepatocellular Carcinoma and Other Liver‐Related Events in Chronic Hepatitis B Patients With Metabolic Dysfunction or Metabolic Dysfunction‐Associated Steatotic Liver Disease

ABSTRACT

Introduction

Metabolic dysfunction and metabolic dysfunction‐associated steatotic liver disease (MASLD) are associated with an increased risk of hepatocellular carcinoma (HCC) in patients with chronic hepatitis B (CHB). We aimed to study risk factors for HCC and to assess the performance of the PAGE‐B score in this population.

Methods

We included CHB patients with ≥ 1 metabolic comorbidity from nine centres. Steatosis was diagnosed by ultrasound, CAP, or histology. Risk factors were analysed by Cox regression, and the performance of the PAGE‐B score was assessed in the overall population and across relevant subgroups.

Results

We included 1922 patients. 1730 (90.0%) were overweight, 434 (22.6%) had hypertension, 254 (13.2%) dyslipidemia, 230 (12.0%) diabetes and 732 (38.1%) MASLD. Presence of cirrhosis, older age, lower platelets and lower albumin were independent risk factors for HCC. The 5‐year HCC risk was 0.1%/2.0%/12.4% patients with low/intermediate/high PAGE‐B scores (p < 0.001). Consistent results were obtained in patients with MASLD (0/2.8/11.1% for low, intermediate and high PAGE‐B scores (p < 0.001)). PAGE‐B stratified risk in patients without cirrhosis (0% vs. 1.2% and 1.8%, p < 0.001). Among the subset of patients with cirrhosis, risks were 4.2% (low), 6.9% (intermediate) and 27.3% (high) (p < 0.001).

Conclusions

CHB patients with metabolic dysfunction and/or MASLD are at significant risk of HCC. The PAGE‐B score can be used to stratify HCC risk in this population, with negligible 5‐year HCC incidence in those without cirrhosis and low PAGE‐B scores. However, caution should be exercised in patients with cirrhosis in whom HCC risk remains significant even among those with a low PAGE‐B score.

This study evaluated the performance of the PAGE‐B score in HCC risk among CHB patients with metabolic dysfunction or MASLD. The PAGE‐B score effectively stratified HCC risk in both MASLD and non‐MASLD patients, especially in those without cirrhosis, where low scores were linked to negligible 5‐year HCC risk.

1. Introduction

Chronic hepatitis B (CHB) infection is a major global health problem. According to the World Health Organization, over 254 million people are living with CHB, and CHB resulted in almost 1.1 million deaths in 2022 [1]. The predominant causes of death among those with CHB are hepatocellular carcinoma (HCC) and complications arising from liver cirrhosis. Studies have highlighted that, in addition to HBV‐related factors and the severity of liver fibrosis, the presence of metabolic comorbidities such as overweight, diabetes mellitus, hypertension, dyslipidemia and metabolic dysfunction‐associated steatotic liver disease (MASLD) is associated with an increased risk of developing HCC and other adverse liver outcomes [2, 3, 4, 5]. In parallel with the metabolic syndrome pandemic, the prevalence of metabolic dysfunction is increasing in individuals with CHB [5]. Given the excess risk of adverse liver‐related outcomes in this population affected by two concomitant liver diseases, effective risk stratification is of major clinical importance.

Previous studies have shown that a combination of patient demographics and biomarkers can be used to stratify HCC risk in patients with CHB. Among these, the PAGE‐B score—which incorporates age, sex and platelet count—has been widely validated for predicting HCC and other liver‐related events in individuals with CHB, including those with hepatitis delta co‐infection [6, 7, 8, 9]. Current guidelines, including the EASL guideline, therefore recommend the use of these scores to identify patients who may or may not require HCC surveillance [10]. However, it currently remains unclear whether these scores can also be effectively applied in patients with CHB co‐affected by metabolic comorbidities, as the presence of these metabolic comorbidities is associated with an increased risk of HCC independent from the aforementioned risk score components [4, 11].

The aim of this study was therefore to explore risk factors for HCC and adverse liver‐related outcomes in patients with CHB with metabolic dysfunction or MASLD, and to study whether the PAGE‐B score can be applied for risk stratification in this population.

2. Methods

2.1. Study Population

For this study we used data from a combined dataset comprising CHB patients from 6 centres in the Netherlands, and from King's College (London, UK), Hospital Universitary Vall d'Hebron (Barcelona, Spain) and TCLD (Toronto, Canada). A summary of the characteristics and enrolment criteria can be found in the Table S1 and previous publications [4, 8, 12].

For the current analysis, we included all patients with CHB mono‐infection who were co‐affected by at least one metabolic comorbidity (overweight, diabetes mellitus, hypertension or dyslipidemia), regardless of the presence of steatosis. We excluded patients in case of known (past) viral co‐infections (HDV, HCV, HIV) or presence of other known chronic liver diseases (documented alcoholic liver disease/alcohol abuse, Wilson's disease, auto‐immune hepatitis, primary biliary cholangitis, primary sclerosing cholangitis and hemochromatosis), and cases in which HCC was diagnosed within 3 months of starting follow‐up.

2.2. Assessment of Metabolic Dysfunction

Presence of metabolic dysfunction was ascertained through chart review. We assessed the presence of diabetes mellitus type 2 (DM2), which was based on medical history or use of antidiabetic medication; hypertension, which was based on medical history or use of antihypertensives; dyslipidemia, which was based on medical history or statin use; and overweight, which was defined as BMI ≥ 25 kg/m² for non‐Asians and ≥ 23 kg/m² for Asians [13]. MASLD was defined as the presence of steatosis (based on histology, controlled attenuation parameter ≥ 248 dB/m and/or ultrasound or CT) in combination with at least one metabolic dysfunction (diabetes, overweight, dyslipidemia or hypertension).

2.3. Assessment of Severity of Liver Disease, and Calculation of the PAGE‐B Score

Presence of cirrhosis was based on liver biopsy showing METAVIR F4 or a liver stiffness measurement (LSM) ≥ 12.0 kPa [14]. In patients without available data on histology and LSM, cirrhosis could be ruled in based on ultrasound findings compatible with cirrhosis and/or portal hypertension (e.g., nodular liver, signs of portal hypertension on imaging or endoscopy). The PAGE‐B score, which is calculated using baseline age, sex and platelet count, is an index that can be used to stratify the risk of HCC in individuals with CHB. The PAGE‐B score was calculated as reported in the original article by Papatheodoridis et al. and individuals were classified as low, intermediate and high risk based on previously reported cut‐offs (< 10, 10–17, > 17) [6].

2.4. Statistical Analysis

The primary outcome of the study was incident HCC, which was diagnosed based on established imaging criteria and/or histopathology. Secondary analyses focused on the incidence of liver‐related events, which were defined as the first occurrence of any of the following: HCC, liver transplantation, or liver‐related death.

Cohort characteristics were described as counts with percentages for categorical variables, as means with the standard deviation (SD) for normally distributed continuous variables, and as medians with the interquartile range (IQR) for non‐normally distributed continuous variables. Differences in characteristics across subgroups were compared using the Kruskal–Wallis test for continuous data and the chi‐square test for categorical data. The cumulative incidence of HCC and liver‐related events were assessed using life‐table methods and the Kaplan–Meier estimator. The follow‐up time for each patient was calculated from baseline date until date of HCC diagnosis, date of transplantation, date of death, or last visit date. Baseline date was set on the date of the first evaluation at the outpatient department after a positive HBsAg test. Analyses were performed in the overall study population and stratified by presence of cirrhosis. Next, we performed Cox regression analysis to assess predictors of HCC and liver‐related events. Factors associated with the outcome of interest (age, sex, ethnicity, platelet count, albumin, ALT, HBV DNA, HBeAg positivity, cirrhosis, antiviral therapy, presence of metabolic dysfunction and steatosis) in univariable analysis were entered into the multivariable model. Finally, we compared the cumulative incidences of HCC and liver‐related events across PAGE‐B risk groups by the log‐rank test. To confirm the robustness of the PAGE‐B scores for risk stratification, additional sensitivity analyses were performed stratified for presence of MASLD and for presence of cirrhosis. Finally, we also assessed the performance of the mPAGE‐B (calculated based on: 1755), CAMD (calculated based on: 1900) and aMAP (calculated based on: 1763) risk scores in this population [11, 15, 16]. Differences were considered statistically significant when p < 0.05. Statistical analyses were performed using SPSS 28.0.1.0 (IBM).

3. Ethics

This study was conducted in accordance with the guidelines of the Declaration of Helsinki and the principles of Good Clinical Practice. The requirement for informed consent was waived (MEC‐2020‐0699, MEC‐2021‐0782 and MEC 2021–0919), and the individual institutional review boards gave approval when necessary.

4. Results

4.1. Patient Characteristics

From the 3628 CHB patients originally in the database, we included 1922 patients with at least one metabolic comorbidity. The median age at baseline was 42 years (interquartile range [IQR] 34–51). The majority of the patients were male (64.7%) and born in sub‐Saharan Africa (41.6%). Within the cohort, 1730 patients (90.0%) were classified as overweight, 434 patients (22.6%) had hypertension, 254 patients (13.2%) had dyslipidemia and 230 patients (12.0%) had diabetes. In total, 732 patients (38.1%) were diagnosed with hepatic steatosis (n = 298 based on histology, n = 144 based on CAP value and n = 290 based on imaging), meeting the criteria for MASLD. Detailed cohort characteristics are described in Table 1.

TABLE 1.

Baseline characteristics.

	Overall cohort n = 1922
Age, years	42 (34–51)
Male sex, n (%)	1244 (64.7)
Ethnicity, n (%)
Caucasian	518 (27.0)
Asian	503 (26.2)
SSA	799 (41.6)
Other	102 (5.3)
ALT, IU/mL	37 (24–62)
Albumin, g/l	45 (42–48)
Platelet count, x109/L	214 (174–255)
HBV DNA, log10 IU/mL	3.4 (2.3–5.3)
HBeAg positive, n (%)	360 (18.7)
Antiviral therapy with NUC, n (%)	1053 (54.8)
Overweight, n (%)	1730 (90.0)
Diabetes, n (%)	230 (12.0)
Hypertension, n (%)	434 (22.6)
Dyslipidemia, n (%)	254 (13.2)
MASLD, n (%)	732 (38.1)
Cirrhosis, n (%)	238 (12.4)

Note: Data are presented as median (IQR) and as n (%) for categorical variables.

The median PAGE‐B risk score was 10 (IQR 6–14), and 873 (45.4%), 830 (43.2%) and 219 (11.4%) patients were classified as low, intermediate and high risk based on their PAGE‐B score. As expected, patients with high PAGE‐B scores were older, predominantly male and showed lower platelet counts. Moreover, they were more likely to have elevated alanine aminotransferase (ALT) levels, reduced albumin levels, higher HBV DNA levels, and were more frequently diagnosed with cirrhosis. Notably, they also had a significantly higher prevalence of diabetes mellitus, hypertension, dyslipidemia and MASLD (Table S2).

4.2. Factors Influencing Hepatocellular Carcinoma Development

During a median follow‐up of 7.9 years (IQR 4.4–11.1 years), HCC was diagnosed in 87 patients. The 5‐ and 10‐year cumulative incidences of HCC were 2.4% (95% confidence interval [CI] 1.6–3.2) and 5.0% (95% CI 3.8–6.2, Figure 1A). Among patients without cirrhosis (n = 1662), the 5‐year and 10‐year cumulative incidence rates were 0.6% and 1.9%, compared to 14.9% and 27.7% among patients with cirrhosis at baseline (n = 238) (Figure 1B, p < 0.001).

FIGURE 1.

Cumulative incidences of HCC (A) in the overall cohort (B) in patients with and without cirrhosis. Patients without available information on fibrosis status at baseline were excluded from the analysis in (B).

Multivariable Cox regression showed that older age (adjusted hazard rate [aHR] 1.04, 95% CI 1.02–1.07, p < 0.001), lower platelet count (aHR 0.99, 95% CI 0.987–0.995, p < 0.001), lower albumin (aHR 0.95, 95% CI 0.98–0.99, p = 0.015), presence of cirrhosis (aHR 3.71, 95% CI 2.12–6.48, p < 0.001) and use of antiviral therapy (aHR 2.74, 95% CI 1.15–6.54, p = 0.023) at baseline were independently associated with HCC development, whereas the presence of steatosis (aHR 0.88, 95% CI 0.56–1.40, p = 0.613) was not (Table 2). We also performed this analysis for patients with and without MASLD and found similar results (Table S3).

TABLE 2.

Baseline variables associated with HCC development.

HCC	Univariate			Multivariate
	HR	95% CI	p	aHR	95% CI	p
Age, years	1.077	1.058–1.096	< 0.001	1.049	1.029–1.070	< 0.001
Male sex	2.240	1.299–3.861	0.004	1.256	0.712–2.216	0.432
Platelet count	0.980	0.977–0.984	< 0.001	0.991	0.987–0.995	< 0.001
Albumin, g/L	0.893	0.866–0.921	< 0.001	0.952	0.915–0.991	0.015
HBV DNA log IU/mL	1.102	1.005–1.202	0.038	0.992	0.906–1.085	0.856
Steatosis	0.961	0.656–1.558	0.961	0.888	0.560–1.407	0.613
Cirrhosis	14.754	9.454–23.025	< 0.001	3.713	2.126–6.484	< 0.001
Antiviral therapy	6.481	2.987–14.059	< 0.001	2.746	1.153–6.540	0.023

Note: Results were obtained with Cox proportional hazard analysis and given as (a)HR with 95% and p‐value.

4.3. PAGE‐B Score Predicts HCC in Patients With Chronic Hepatitis B and Metabolic Dysfunction and MASLD

The 5‐ and 10‐year cumulative incidence of HCC in the low risk group was 0.1% and 0.7%, compared to 2.0% and 4.8% in the intermediate and 12.4% and 20.8% in the high risk groups (Figure 2, p < 0.001). Consistent results were obtained for patients with diabetes, hypertension, dyslipidaemia, and/or with overweight (all p < 0.01, Figure S1A–D) and findings were also consistent regardless of the presence of MASLD. The 5‐ and 10‐year cumulative HCC incidences for patients without MASLD were 0.2% and 1%, 1.5% and 5.4% and 13.4% and 23.1% for those with low, intermediate and high PAGE‐B scores (Figure 3A, p < 0.001) and 0% and 0%, 2.8% and 4.0% and 11.1% and 17.5% for patients with MASLD (Figure 3B, p < 0.001).

FIGURE 2.

Cumulative incidences of HCC according to PAGE‐B strata.

FIGURE 3.

Cumulative incidences of HCC according to PAGE‐B score in patients without (A) and in patients with (B) MASLD.

The PAGE‐B score effectively stratified the risk of HCC among patients without cirrhosis (n = 1662) at baseline; the 5‐year cumulative incidences were 0%, 1.2% and 1.8% for patients with low, intermediate and high risk PAGE‐B scores (Figure 4A, p < 0.001). Higher PAGE‐B scores were also associated with a higher risk of HCC among patients with cirrhosis (n = 238): the 5‐year cumulative incidences were 4.2%, 6.9% and 27.3% for patients with low, intermediate and high PAGE‐B scores (Figure 4B, p < 0.001).

FIGURE 4.

Cumulative incidences of HCC according to PAGE‐B score in patients without (A) and in patients with (B) cirrhosis.

In a sensitivity analysis only including patients on NUC treatment during follow‐up, we found consistent results: the cumulative incidence of HCC at 5 years was 0.3%, 2.6% and 14.0% for low, intermediate and high‐risk PAGE‐B scores (Figure S2, p < 0.001). Findings were consistent for mPAGE‐B, CAMD and aMAP scores (Table S4).

4.4. PAGE‐B Showed Consistent Results for Predicting Liver‐Related Events

During the same median follow‐up period of 7.9 years (IQR 4.4–11.1 years), 108 patients experienced a first liver‐related event, comprising 87 HCCs, 12 liver transplantations and 9 liver‐related deaths (not HCC related). The 5‐ and 10‐year cumulative incidence of liver‐related events was 2.9% (95% CI 2.1–3.6) and 6.1% (95% CI 4.7–7.4).

The 5–10 years cumulative incidence of liver‐related events in the low risk PAGE‐B group was 0.1% and 0.7%, compared to 2.7% and 5.7% in the intermediate risk group and 14.7% and 26.4% in the high risk group (Figure S3, p < 0.001). Consistent results were obtained across patients with diabetes, hypertension, dyslipidaemia, overweight (all p < 0.01, Figure S4A–D). Findings were also consistent regardless of the presence of MASLD. The 5 year cumulative incidences of liver‐related events for patients without MASLD were 0.2%, 1.9% and 16.8% for patients with low, intermediate and high PAGE‐B scores (Figure S5A, p < 0.001) and 0%, 3.4% and 12.1% for patients with MASLD with low, intermediate and high PAGE‐B scores (Figure S5B, p < 0.001).

5. Discussion

In this large multicenter study comprising 1922 CHB patients with metabolic dysfunction, we identified older age, lower platelet count, lower albumin levels, and the presence of cirrhosis as risk factors for HCC development. Second, the PAGE‐B score, which uses older age and platelet count, was highly predictive of HCC. This confirms that the PAGE‐B score, similar to its use in other patient populations, can also be used to guide HCC surveillance strategies in patients with CHB and metabolic dysfunction, although caution should be exercised in patients with cirrhosis as a risk of HCC persists even in those with a low PAGE‐B score.

Multiple studies have demonstrated an increased risk of adverse liver outcomes in CHB patients with metabolic comorbidities [3, 4, 5]. This may be attributable to more rapid progression of liver fibrosis in this population, as recently demonstrated in studies of serial liver stiffness assessments [12, 17]. Given the high background risk of HCC in patients with CHB, enrolment in an HCC surveillance programme is advised for patients with cirrhosis, but also for patients at excess risk based on Asian or Sub‐Saharan African ethnicity, male sex and older age [18]. However, the risk of HCC may vary widely within the population currently considered eligible for HCC surveillance, suggesting that personalised risk assessment rather than a one‐size‐fits‐all approach is warranted. Various risk scores have been developed to stratify HCC risk among CHB patients, with the PAGE‐B score being one of the most widely validated [6]. This score, designed for the 5‐year prediction of HCC in CHB patients, has been validated in several patient populations with CHB with or without HDV or HIV co‐infection [7, 8, 19, 20]. Based on these studies, the EASL guideline already suggests that HCC screening may be foregone in some CHB patients with a low PAGE‐B score [21]. However, previous studies indicate that the presence of metabolic dysfunction or MASLD is a risk factor for adverse liver outcomes independent from the factors included in the PAGE‐B score [9]. Whether PAGE‐B can be effectively applied in CHB patients co‐affected by metabolic dysfunction, therefore, remained unclear.

In the current study, higher PAGE‐B scores were associated with a higher risk of HCC. The 5‐year cumulative incidence was only 0.1% in the low PAGE‐B score group, compared to 2.0% and 12.4% in the intermediate and high PAGE‐B score groups, and these numbers are comparable to the PAGE‐B derivation cohort [6]. The observation that PAGE‐B effectively stratifies HCC risk in CHB with metabolic dysfunction is of major clinical relevance, given the increasing prevalence of metabolic dysfunction in the aging CHB population [22]. Importantly, these findings were consistent across the various different metabolic comorbidities, and across patients with and without MASLD. These findings can potentially be used in HCC surveillance strategies to exempt or enrol patients based on personalised HCC risk. Since recent studies indicate that treatment of metabolic comorbidities and MASLD is associated with reductions in severity of fibrosis and/or HCC risk, the current findings also suggest that the PAGE‐B score could potentially be used as a non‐invasive tool to identify patients more likely to benefit from treatments directed at the co‐existing metabolic comorbidity [23, 24, 25].

Interestingly, patients with metabolic dysfunction who also had steatosis did not have a higher risk of adverse outcomes when compared to the patients with metabolic dysfunction alone. This finding suggests that the elevated HCC risk observed in CHB patients with MASLD may be primarily attributable to the underlying metabolic dysfunction, rather than steatosis itself. This finding is in line with a recent report that showed that metabolic dysfunction was associated with more severe histological lesions in patients with CHB treated with antivirals, whereas the concomitant presence of steatosis was not [26]. Furthermore, various other recent studies showed that steatosis was not associated with an increased risk of adverse outcomes when adjusted for the presence of metabolic comorbidities [27, 28, 29].

In addition to the prediction of HCC, risk stratification for other adverse liver‐related events is also of great importance, as this risk is also elevated in patients with CHB who are co‐affected by metabolic dysfunction. In line with our observations regarding HCC, we also observed a higher risk of liver‐related events in patients with higher PAGE‐B scores, even though most of the liver‐related events in our cohort were based on HCC development. This is most likely due to the fact that, in the current era of effective antiviral therapies, few patients develop decompensated liver disease while on antiviral treatment [30, 31].

Despite the excellent performance in the overall cohort, PAGE‐B may not be optimal in patients with established cirrhosis. In this subset of patients, the PAGE‐B score discriminated low versus intermediate and high risk of HCC, but a substantial risk of HCC remained in the predicted low risk group (4.2% at 5 years). This persistent risk of HCC was also observed in the subset of patients with a low predicted risk based on alternative scores such as mPAGE‐B and aMAP. Although this incidence is not negligible, it remains to be determined whether this is sufficiently high to warrant HCC surveillance. The newly updated European Association for the Study of the Liver (EASL) guidelines on HCC recommend a surveillance threshold of 1.5% annual risk in patients with cirrhosis as the lower limit for cost‐effectiveness, although this may vary widely across countries and willingness to pay perspectives [32]. Nevertheless, our findings highlight that the PAGE‐B score has important prognostic value in this population, for HCC but also for other liver‐related events.

In addition to the components of the PAGE‐B score, albumin levels and the use of antiviral therapy were found to be associated with the risk of HCC. A modified PAGE‐B (mPAGE‐B) risk score, which incorporates albumin, was already proposed by Kim et al. and subsequently validated in an Asian cohort of patients receiving antiviral therapy [15]. In our cohort, we calculated the mPAGE‐B score whenever albumin data was available and observed similar cumulative incidence rates as those with the PAGE‐B score. The 5‐year cumulative incidence of HCC was 0.1% for the low mPAGE‐B score group, 2.8% for the intermediate mPAGE‐B score group, and 14.2% for the high mPAGE‐B score group. More comparable HCC risk scores exist that use one or more variables similar to those in the PAGE‐B score. Using available data, we calculated these scores for our cohort where possible and determined the 5‐year cumulative incidence of HCC for the overall cohort as well as stratified by cirrhosis status. These findings are summarised in the Table S4. Given the comparable performance, we feel the use of PAGE‐B might be preferred since it is simpler.

A somewhat counterintuitive finding in our analysis was the association between treatment with antivirals and an increased risk of adverse outcomes. This is probably attributable to confounding by indication: patients who are eligible for antiviral treatment generally have more advanced disease, characterised by higher ALT levels, elevated HBV DNA, and greater degrees of liver fibrosis, and are therefore at higher risk for developing HCC.

Despite the fact that this is a large multicenter cohort only including CHB patients with metabolic dysfunction, there are some limitations. First, this is a retrospective cohort study and detailed information on the presence of other HCC‐related risk factors such as the presence of moderate alcohol use, family history of HCC, and other lifestyle factors was not always available. Also, the presence of metabolic dysfunction was based on chart review and we were not able to account for treatment effect. Furthermore, 81 patients were already on antiviral therapy before the baseline date, which could influence some measurements. Exclusion of these patients did not influence any of the findings. Additionally, the majority of patients with metabolic dysfunction in this cohort were overweight, whereas other metabolic comorbidities were less often observed. We therefore assessed the performance of the PAGE‐B score for every metabolic comorbidity separately and found similar results (Figure S4). Additionally, the presence of steatosis was determined based on results obtained with biopsy, CAP, or imaging. Since the diagnostic accuracy of CAP is limited, ultrasound has limited sensitivity for non‐severe steatosis, and since steatosis may resolve, it is more difficult to detect in patients with cirrhosis; steatosis may have been underdiagnosed. In 16 patients, the presence of steatosis could not be assessed since there was no histological or imaging data available. This limitation is unlikely to have impacted the study results, as only one liver‐related adverse event was observed in this subgroup. Third, liver stiffness thresholds for cirrhosis remain debated and vary across studies. We used a cut‐off of 12 kPa for cirrhosis, a threshold commonly applied in patients with chronic hepatitis B. However, some guidelines recommend using a higher cut‐off in patients with MASLD. We therefore performed a sensitivity analysis increasing the cut‐off to ≥ 16 kPa, as suggested by the AASLD Practice guideline on imaging‐based noninvasive assessment of hepatic fibrosis and steatosis. With this threshold, 217 patients were classified as having cirrhosis, compared to 238 using the 12 kPa. Findings were consistent in this subgroup. Lastly, we assessed metabolic dysfunction and steatosis at baseline and there was not always data on the development of metabolic dysfunction during follow‐up.

In conclusion, patients with CHB co‐affected by metabolic dysfunction have a significant risk of HCC. Older age, lower platelet count, lower albumin levels and presence of cirrhosis are risk factors for the development of HCC, and PAGE‐B effectively stratified for HCC risk. A low PAGE‐B score is associated with a negligible risk of HCC in patients without cirrhosis and can be used to identify patients in whom HCC surveillance can be safely discontinued. Among patients with established cirrhosis, a residual risk of HCC remains even among patients with low PAGE‐B scores, underscoring the need for continued HCC surveillance.

Author Contributions

Lesley A. Patmore: conceptualization, investigation, writing – original draft, methodology, validation, visualization, formal analysis, project administration, writing – review and editing. Ivana Carey: data curation, writing – review and editing. Jordan J. Feld: data curation, writing – review and editing. Willem P. Brouwer: writing – review and editing. Keyur Patel: writing – review and editing. Maria Buti: writing – review and editing, data curation. Pieter Honkoop: data curation, writing – review and editing. Douwe F. Postma: writing – review and editing. Hans Blokzijl: writing – review and editing, data curation. Özgur. M. Koc: data curation, writing – review and editing. Eva van Oorschot: writing – review and editing. Kosh Agarwal: writing – review and editing. Marc van der Valk: writing – review and editing, data curation. Faydra I. Lieveld: writing – review and editing, data curation. Mai Kilany: writing – review and editing, data curation. Matthijs Kramer: writing – review and editing. Joep de Bruijne: writing – review and editing. Mark A. A. Claassen: writing – review and editing. Bettina E. Hansen: writing – review and editing. Robert A. de Man: writing – review and editing. Harry L. A. Janssen: writing – review and editing. R. Bart Takkenberg: writing – review and editing, data curation. Milan J. Sonneveld: writing – review and editing, supervision, funding acquisition, conceptualization, writing – original draft, methodology, formal analysis.

Ethics Statement

This study was conducted in accordance with the guidelines of the Declaration of Helsinki and the principles of Good Clinical Practice. The requirement for informed consent was waived, and the individual institutional review boards gave necessary approval. The study protocols were reviewed by the Erasmus MC Medical Ethical Committee.

Conflicts of Interest

L.A.P.: research support and consultancy fee from Gilead Sciences. I.C.: Received a research grant from Gilead. J.J.F.: has received personal fees for consulting from AbbVie, Enanta, Gilead, Janssen and Roche, and research funding from AbbVie, Abbott, Gilead and FUJIFILM Wako Chemicals. W.P.B.: received a speakers fee from Eli Lilly. Is on the advisory board of Novo Nordisk. Participated in trials from Inventive pharma, Boehringer Ingelheim and 89BIO. K.P.: DSMB for Gilead Sciences, consulting for Novo‐Nordisk, Resalis, Merck. M.B.: Received grants and Advisory Board Gilead Sciences. P.H.: none. D.F.P.: Member of DSMB of the COBRA‐trial (Very short‐course versus standard course antibiotic therapy in patients with acute ChOlangitis after adequate endoscopic BiliaRy drAinage (COBRA trial)); consultancy fees from Gilead (payed to institution). H.B.: none. O.M.K.: none. E.vO.: none. K.A.: received a research grant from Gilead, Roche and MSD. Is an advisor for Aligos, Arbutus, Assembly, Abbvie, Biotest, Bluejay, GLG, Gilead, Immunocore, Shiniogi and Vir Biotechnology Inc. M.vd.V.: Unrestricted research funding and consultancy fees from Gilead, MSD, ViiV, all paid to his institution. F.I.L.: none. M.Ki.: none. M.Kr.: received a speakers fee from Norgine. J.dB.: research support from Terumo. M.A.A.C.: none. B.E.H.: received grants from Mirum Pharma, Intercept, Cymabay, Eiger, Ipsen/Albireo, Pliant, Advanz and Calliditas. Is a consultant for Mirum Pharma, Intercept, Cymabay, Eiger, Pliant, Advanz and Calliditas. R.A.dM.: research support from Roche. H.L.A.J.: received grants from: Gilead Sciences, GlaxoSmithKline, Janssen, Roche, Vir Biotechnology. Consultant for: Aligos, Gilead Sciences, GlaxoSmithKline, Grifols, Roche, Vir Biotechnology Inc., Precision Biosciences. R.B.T.: Speakers fees from WL gore, research support Norgine, consultancy fee 89Bio. M.J.S.: Speakers fees and research support from Gilead, Roche, Fujirebio and consultancy fees from Gilead and Albireo.

Supporting information

Figure S1: Cumulative incidences of HCC in (A) patients with diabetes according to PAGE‐B strata, (B) patients with hypertension according to PAGE‐B strata, (C) patients with dyslipidemia according to PAGE‐B strata and (D) patients with overweight according to PAGE‐b strata.

Figure S2: Cumulative incidences HCC in patients with NUC according to PAGE‐B strata.

Figure S3: Cumulative incidence of liver‐related events according to PAGE‐B strata.

Figure S4: Cumulative incidences of liver‐related events according to PAGE‐B score for patients with (A) diabetes, (B) hypertenstion, (C) dyslipidemia and (D) overweight.

Figure S5: Cumulative incidences of liver‐related events according to PAGE‐B score for patients without (A) and patients with (B) MASLD.

Table S1: Characteristics of included cohorts.

Table S2: Baseline characteristics of patients according to PAGE‐B score. Data are presented as median (IQR) and as n (%) for categorical variables.

Table S3: Baseline variables associated with HCC development in patients with and without MASLD. Results were obtained with Cox proportional hazard analysis and given as (a)HR with 95% and p‐value.

Table S4: 5‐year cumulative incidence according to different HCC risk scores categories for the overall cohort and stratified by presence of cirrhosis.

Data Availability Statement

The data that support these findings are not publicly available, since they are subject to (inter)national data protection laws to ensure data privacy of the study participants. The data can therefore not be shared.