Temporal evolution of liver cancer etiology in China, 1990–2021: insights from the Global Burden of Disease Study 2021

Introduction

Liver cancer ranks as the fifth most prevalent malignant neoplasm and is the third most common cause of cancer-related mortality globally. As per the most recent data from the International Agency for Research on Cancer (IARC), in 2022, there would be 866,100 new liver cancer cases and 758,700 fatalities globally, with China representing 42.5% and 41.7% of these totals, respectively. In China, the incidence of liver cancer is the fifth highest among all kinds of cancer, while its fatality rate is the second highest (1-6). Primary liver cancer is classified into pathological kinds: hepatocellular carcinoma (HCC), intrahepatic cholangiocarcinoma (ICC), and mixed varieties, with HCC being the predominant form, constituting 93% of all liver malignancies in China (7,8). The hepatitis B virus (HBV) infection is the predominant cause of liver cancer in China, responsible for 84.4% of all liver cancer cases, with notably elevated frequency in the southeastern coastal regions and rural high-incidence areas inland. These countries generally exhibit warm, humid temperatures that facilitate aflatoxin growth in cereals like corn and peanuts. Aflatoxins are recognized as powerful carcinogens that demonstrate synergistic carcinogenic effects in conjunction with HBV infection. The hepatitis C virus (HCV) infection prevalence is low at 3.2%, although its contribution has been progressively rising in recent years due to population mobility and an increased risk of medical exposure (9-14). Concurrently, as urbanization intensifies, the incidence of alcohol-related liver disease and non-alcoholic steatohepatitis (NASH) has persistently increased in developed areas (15-19). Aflatoxin, a critical risk factor for liver cancer, significantly interacts with HBV infection, particularly in southern provinces characterized by hot and humid climates. Its compounded effect on metabolic diseases, such as obesity and diabetes mellitus, exacerbates the proliferation of the high-risk population (20).

China has achieved significant advancements in the primary prevention of liver cancer: the neonatal HBV vaccination coverage exceeds 99%, resulting in a reduction of the HBV carriage rate among adolescents to below 1%; additionally, the widespread use of HCV direct antiviral drugs (DAAs) has markedly diminished the risk of related liver cancer. The prevalence of NASH-related liver cancer has escalated rapidly, indicating that the adverse impacts of dietary modifications and sedentary habits are becoming significant (7,8,21).

This study systematically examines the spatial and temporal characteristics of the liver cancer burden, as well as that attributable to the five principal etiologies (HBV, HCV, alcohol consumption, NASH, and other causes), based on Global Burden of Disease (GBD) 2021 data, and analyzes the correlation between social development levels and the disease burden index. The study analyzes the primary factors exacerbating the disease burden to establish a scientific foundation for China to develop a targeted intervention strategy for liver cancer based on “graded prevention and control and localized measures”. We present this article in accordance with the STROBE reporting checklist (available at https://cco.amegroups.com/article/view/10.21037/cco-25-80/rc).

Methods

Data sources

This study was based on the GBD 2021 database, which covers 204 countries and regions and 371 diseases and injuries globally. The study extracted data on incidence, mortality, and disability-adjusted life year (DALY) of liver cancer and liver cancer due to five major causes in China from the Global Health Data Interactive (GHDx) website (https://vizhub.healthdata.org/gbd-results/) for the period of 1990–2021, and was stratified by region, age, and gender in China. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Data analysis and processing

Observational indicators and definitions

• Crude rate = number of cases (incidence, deaths, DALY) in a given year/number of people at risk in the same period ×100,000
• Age-standardized rate (ASR): weighted by world standard population structure:
  $\text{ASR}=\frac{\text{Crude}\text{rate}\text{of}\text{each}\text{age}\text{group}\times \text{standard}\text{population}\text{weight}}{{\displaystyle \sum \text{standard}\text{population}\text{weight}}}$
• DALY = years of life lost (YLL) + years of disability lost (YLD)
• YLL = standardized life expectancy − actual age at death
• YLD = number of cases × disease disability weight × average duration of disability
• Socio-Demographic Index (SDI): a composite indicator [0–1] measuring the level of social development of countries and regions.

Analysis tools

The study utilized R 4.4.2 for statistical analysis and visualization, Joinpoint software (5.0.2) for Joinpoint regression analysis, the BAPC package and forecast package for projection analysis.

Statistical analysis

• Joinpoint regression analysis: to precisely identify the time points at which significant changes occurred in the trends of China’s liver cancer disease burden between 1990 and 2021 this model employs a log-linear regression framework. By introducing connection points, it segments the time series data into multiple consecutive intervals and performs independent trend fitting within each interval. In model specification, the optimal number and positions of connection points, along with their corresponding regression coefficients, were determined using a “grid search” approach combined with a “Monte Carlo permutation test”. The permutation count was set to 4,499 with an overall significance level (α) of 0.05. Model selection followed the principle of minimizing the Bayesian Information Criterion (BIC) to ensure an optimal balance between model fit quality and complexity. The final regression model is expressed as: E[y|x] = e^(β₀ + β₁x + δ₁(x-τ₁)₊ + ... + δₖ(x-τₖ) ₊), where y represents the age-standardized rate, x denotes the calendar year, τₖ is the kth connection point, β and δₖ are regression coefficients, and the operator a₊ equals a when a >0, otherwise it equals 0.
  Model results are interpreted using the following two core metrics:
  • Annual percentage change (APC): quantifies the annual rate of change within each independent trend segment. Its point estimate is calculated as ‘APC = (e^β₁ − 1) × 100%’, with its 95% confidence interval reported.
  • Average annual percentage change (AAPC): as a summary statistic for the entire study period [1990–2019], AAPC is the weighted average of segment APC weighted by their interval lengths, also reporting its 95% confidence interval (22).
• The APC model was constructed to analyze the impact of three different time-related factors (age, period, and cohort) on disease burden status. Age effects represent risk variations due to biological aging; period effects capture influences from temporal factors affecting all age groups simultaneously; cohort effects reflect health impacts from shared early-life exposures among individuals born in the same era.
  Data were organized into 5-year intervals for ages 15–89 years and periods 1994–2021, generating birth cohorts from 1905–1909 to 2000–2004. We fitted the log-linear model: ln(Yₐₚ) = μ + αₐ + βₚ + γₜ + ε, where Yₐₚ represents age-standardized rates for age group a in period p, with cohort t derived as t = p − a.
  The intrinsic estimator method addressed the identifiability problem from the perfect linear relationship among the three dimensions. Results are presented as relative risks using mean values as reference (23).
• Health inequality analysis: we quantified socioeconomic inequalities in liver cancer burden using two standardized measures recommended for health inequality monitoring. The analysis utilized age-standardized DALY rates to ensure comparability across populations with varying age structures.
  The SII was employed to measure absolute health inequality. This index was derived from a weighted regression model that establishes the relationship between age-standardized DALY rates and the relative socioeconomic position of each region, ranked according to the SDI. The SII represents the absolute difference in health outcomes between the most and least advantaged groups in the population.
  The Concentration Index (CI) was used to assess relative inequality. This index measures the extent to which liver cancer burden is concentrated among specific socioeconomic groups by comparing the cumulative distribution of disease burden against the cumulative distribution of the population, ranked by socioeconomic status. The CI ranges from −1 to 1, where negative values indicate that the disease burden disproportionately affects socioeconomically disadvantaged populations, while positive values suggest concentration among more advantaged groups (24).
• Autoregressive Integrated Moving Average (ARIMA) modeling: we employed the ARIMA framework to forecast future trends in liver cancer burden. The ARIMA model integrates autoregressive (AR) and moving average (MA) components to capture temporal dependencies in time series data. The general model structure is defined as: Yt = c + ϕ1Yt-1 + ... + ϕpYt-p + θ1εt-1 + ... + θqεt-q + εt
  where Yt represents the observed value at time t, ϕi are autoregressive parameters, θj are moving average parameters, εt is white noise error term, and p and q denote the orders of AR and MA components, respectively.
  Final model selection was based on minimizing the Akaike Information Criterion (AIC). Model adequacy was verified through residual diagnostics, including the Ljung-Box test for white noise and examination of residual autocorrelation patterns (22).
• Bayesian Age-Period-Cohort (BAPC) modeling: to complement the ARIMA forecasts and account for demographic influences, we implemented a BAPC model. This approach extends the classical APC framework through Bayesian inference, providing robust uncertainty quantification for future projections. The BAPC model was specified as:
  Yapt ~ Poisson(μapt)
  log(µapt) = αa + βp + γt + εapt

Where Yapt represents disease counts for age group a, period p, and cohort t, with αa, βp, and γt representing age, period, and cohort effects respectively. We employed random walk priors of order 1 or 2 for the temporal components to ensure smoothness across adjacent age groups, periods, and cohorts. Model estimation was performed using Markov Chain Monte Carlo (MCMC) methods implemented in the R ‘BAPC’ package, with convergence assessed through Gelman-Rubin statistics and trace plot examination.

Forecasts were generated for the period 2022–2036 by projecting the estimated period and cohort effects while maintaining the observed age structure. Results are presented as posterior medians with 95% credible intervals, reflecting the full uncertainty in parameter estimation and future projection (25).

Results

Incidence, mortality, DALY burden

In 2021, the global burden of liver cancer exhibited considerable geographic variability, with its etiological composition closely linked to regional public health conditions. East Asia, West Africa, and Southeast Asia were identified as high-prevalence regions, likely due to the elevated incidence of HBV (Figures S1-S3). From both Chinese and global viewpoints, HBV and HCV predominated among the different etiological factors. Between 1990 and 2021, the age-standardized incidence rate (ASIR) of liver cancer in China exhibited a marginal decline [estimated annual percentage change (EAPC) =−0.31], but the absolute number of cases escalated markedly from 96,434 to 196,637, reflecting a percentage shift of 103.91%. The global ASIR exhibited a modest increase (EAPC =0.04), whereas the number of cases surged significantly (116.28%); the age-standardized mortality rate (ASMR) and age-standardized disability-adjusted life years rate [AS(DALY)R] for the disease demonstrated an overall declining trend (EAPC =−0.93 and −0.16, respectively), with a more pronounced reduction observed in China compared to the global average (Tables 1,2). The ASIR of HBV-related liver cancer declined in both China (EAPC =−0.55) and globally (EAPC =−0.31), primarily due to a reduction in incidence over the past five years; however, the absolute number of incidence cases rose from 63,118 to 118,665 in China and from 109,479 to 206,366 globally. Liver cancer attributable to HCV exhibited analogous changes in China to those caused by HBV, but with a lesser degree of alteration compared to the latter. In comparison to 1990, the ASIR of NASH-related liver cancer in China exhibited a general increase in 2021 (EAPC =0.73 for China, 0.99 globally); conversely, the ASMR and AS(DALY)R both experienced a decline (EAPC: −0.15 and −0.16, respectively). The age-standardized rate for alcohol-related liver cancer burden exhibited a pattern analogous to that of NASH-related liver cancer. Globally, the age-standardized rate for liver cancer burden linked to metabolic variables exhibited an increasing trend. The ASIR, ASMR, and AS(DALY)R for liver cancer attributed to other causes have declined in China, with EAPC of −0.46, −1.05, and −1.37, respectively (Tables 1,2, Figure 1). A decomposition study indicated that from 1990 to 2021, the DALY for liver cancer and liver cancer attributable to the five principal etiologies in China exhibited a general upward trend. The primary positive contributor to the increase in DALY is aging, exemplified by a rise of up to 141.91% in liver cancer cases. Population expansion positively influenced the increase in DALY, but to a lesser degree than aging, accounting for 52.47% in liver cancer. Epidemiological developments adversely affected the increase of DALY, particularly with viral hepatitis B and other etiologies of liver cancer (Table 3, Figure 2).

Figure 1 Temporal trends in ASIR, mortality rate ASMR, and AS(DALY)R for liver cancer and liver cancer due to 5 major causes (HBV, HCV, NASH, alcohol use, other causes) in China and globally between 1990 and 2021. AS(DALY)R, age-standardized disability-adjusted life year rate; ASIR, age-standardized incidence rate; ASMR, age-standardized mortality rate; HBV, hepatitis B virus; HCV, hepatitis C virus; NASH, non-alcoholic steatohepatitis.

Figure 2 Decomposition analysis of the change in liver cancer DALYs by etiology and sex in China, 1990–2021. The bars quantify the absolute change in DALYs over the study period, decomposed into the contributions of three driving factors: population growth, population aging, and changes in age-specific epidemiological rates. (A) Liver cancer; (B) liver cancer due to hepatitis B; (C) liver cancer due to hepatitis C; (D) liver cancer due to NASH; (E) liver cancer due to alcohol use; (F) liver cancer due to other causes. A positive value indicates an increase in DALYs driven by that factor. DALYs, disability-adjusted life years; NASH, non-alcoholic steatohepatitis.

The disease burden of liver cancer in China exhibits notable age and gender disparities: the incidence of liver cancer and its various etiologies predominantly impacts individuals over 55 years of age, with the ASIR escalating most rapidly among those aged 55–75 years. Conversely, ASMR and AS(DALY)R) decelerate post 75 years, with a marked decline after 90 years. The influence of NASH and alcohol-related liver cancer on healthy life years lost is significantly greater for the 45–70 years age demographic, peaking between 60 and 65 years. In contrast, other liver cancer types predominantly affect a younger population (Figures 3,4). The burden of disease from HCV-associated liver cancer was greater in females than in males, while the burden from all other liver cancer types was higher in males (Figures S4-S11). The age compositions of males and females did not exhibit significant differences (Figures S12-S14). Decomposition analysis indicated that the impact of age on the increase of DALYs was consistently greater in females than in males across all forms of liver cancer. In HBV-related liver cancer, the impact of aging on the increase of DALYs in females was 301.7%, markedly above the 136% observed in males (Figure 2, Table 3). Joinpoint regression analysis revealed a significant decline in the ASIR of liver cancer over the periods of 2000–2005 (APC =−3.39) and 2016–2021 (APC =−1.70). Additionally, the ASMR and AS(DALY)R exhibited a consistent drop from 1990 to 2021 (Figure S15). The ASIR of HBV-associated liver cancer exhibited a fall throughout two intervals (2000–2005, 2016–2021), while the ASMR consistently decreased during the 20-year span (2001–2021). Additionally, the AS(DALY)R diminished across three distinct periods (2001–2005, 2009–2012, 2015–2021) (Figure S16). The characteristics of alterations in HCV-associated liver cancer were analogous to those in HBV-associated liver cancer (Figure S17). The ASIR for liver malignancies associated with NASH increased over a decade [2005–2015] and has subsequently experienced a reduction, with an accelerated rate of decrease in the past three years (APC =−2.85); similar trends are observed in the ASMR and AS(DALY)R (Figure S18). The age-standardized rates of disease burden linked to liver cancer from alcohol consumption were comparable but exhibited a lesser overall rate of change than those associated with liver cancer from NASH (Figure S19). While Joinpoint regression analysis reveals a decreasing trend in the ASIR, ASMR, and AS(DALY)R for liver cancer linked to metabolic diseases over the last decade, this study further illustrates, by examining the disease burden proportions from five prevalent causes of liver cancer between 1990 and 2021, that the overall contribution of liver cancer attributable to metabolic diseases exhibits a notable upward trend (Figure 5). The ASIR of liver cancer from other causes decreased significantly following a substantial rise between 2005 and 2015, although the age-standardized mortality rate continued to decrease generally, albeit with minor oscillations (Figure S20).

Figure 3 Distribution of incidence, mortality, and DALY rates for liver cancer by age group and etiology in China, 2021. DALY, disability-adjusted life year; NASH, non-alcoholic steatohepatitis; Val, value.

Figure 4 This figure illustrates the burden of liver cancer in China in 2021 due to five major causes, distributed by age. (A-C) Chord diagrams showing the mapping between age-standardized rates of different types of liver cancer (per 100,000 people) and nine age groups: (A) ASIR; (B) ASMR; (C) AS(DALY)R. (D-F) Stacked bar charts showing the age composition of disease burden for different types of liver cancer: (D) incidence; (E) deaths; (F) DALY. AS(DALY)R, age-standardized disability-adjusted life year rate; ASIR, age-standardized incidence rate; ASMR, age-standardized mortality rate; DALY, disability-adjusted life year; NASH, non-alcoholic steatohepatitis.

Figure 5 Proportional changes in ASIR, ASMR, and AS(DALY)R for liver cancer by different etiologies, 1990–2021. AS(DALY)R, age-standardized disability-adjusted life year rate; ASIR, age-standardized incidence rate; ASMR, age-standardized mortality rate; NASH, non-alcoholic steatohepatitis.

APC analysis showed that the incidence of liver cancer increased significantly with age, starting slowly in the 20s, increasing sharply above 60 years of age, and peaking at around 80 years of age. The birth cohort effect was significant, with the early cohort around 1880 having a very high risk of incidence (relative risk ratio of about 15), while the modern cohort after 2000 had its risk reduced to close to 0. The period effect was relatively weak, with the incidence rate generally stable between 1990 and 2025, fluctuating between 16 and 17, with relative risk ratios ranging between 0.9 and 1.1. The burden of liver cancer was primarily driven by age and birth cohort; comparing liver cancer overall, the period effect for liver cancer due to HBV and HCV declined and fluctuated, respectively, with the cohort effect higher in the early cohort; liver cancer due to NASH rose over time, with a higher risk in the advanced cohort; and liver cancer due to alcohol intake and other causes was similarly characterized to liver cancer overall (Figures S21-S26).

Relationship between the burden of disease and SDI

The study analyzed the relationship between SDI and different burden of disease indicators in 204 countries and regions in 2021: the correlation curves between age-standardized rates and SDI for liver cancer and liver cancer of different etiologies were downward parabolic in countries and regions with low SDI levels (<0.5), with the peak of the curve around 0.25. The correlation curves for the range of 0.5–0.75 were similar but the peak of the curve near 0.6 was much smaller than the peak of 0.25. The differences are reflected in countries and regions with SDI >0.75, the overall ASIR and ASMR for liver cancer show a slight increase after reaching the lowest point, and then remain at that level, and AS(DALY)R shows a sustained low-level increase; the age-standardized indices of liver cancer caused by HBV and NASH showed a continuous mild decrease, while HCV, alcohol use, and other causes -associated liver cancer showed a low-level and continuous increase after that. China is an upper-middle-income (SDI ≈0.75) country . However, the burden of disease for liver cancer and liver cancer of different etiologies in China is at a high level among countries with the same SDI level, indicating that China is still facing a big challenge to prevent and treat liver cancer (Figures S27-S32). From 1990 to 2021, ASIR, ASMR, and AS(DALY)R of liver cancer and liver cancer due to different etiologies showed a negative trend of correlation with SDI, although the findings were not statistically significant (Figure S33). The health inequality analysis showed that the SII of DALY for liver cancer increased from −28.97 [1990] to 21.57 [2021], indicating a significant increase in absolute health inequality between high-income and low-income countries, with the remaining liver cancers of different etiologies showing similar changes. In 1990, the CI for all six disease types decreased, indicating a reduction in relative inequality (Figures S34,S35).

Future projections

The ARIMA model predictions show that: The ASIR for liver cancer will increase somewhat after 2021, peak around 2029 and then show a downward trend, and basically remain at that level after reaching the lowest point in 2035, with a more pronounced trend in males; the ASIRs for HBV, NASH, and liver cancers due to other causes all show a similar predicted change in ASIR for viral hepatitis B, NASH, and other causes of liver cancer, while alcohol use-associated liver cancer maintained that level after a mild increase and viral hepatitis C-associated liver cancer showed a continued decline; AS(DALY)R for alcohol intake-associated liver cancer maintained that level after a mild increase, and the rest of the disease types were predicted to remain at their current levels through 2036 (Figure 6). In the analysis of liver cancer burden models for the total population, diagnostic plots and statistical indicators collectively reveal the effectiveness and limitations of the models. ACF and PACF plots demonstrate significant autocorrelation characteristics across different etiologies: alcohol-use-related liver cancer incidence exhibits distinct peaks at specific lags (ACF values exceeding confidence intervals), indicating short-term autocorrelation; the PACF plot for hepatitis B-related incidence shows a significant positive correlation at lag 3, suggesting the necessity of incorporating an AR(3) term; while the ACF plot for hepatitis C-related incidence displays rapid decay of autocorrelation coefficients, aligning with moving average features. Concurrently, Q-Q plot analysis reveals that residual distributions for most models approximate normality (e.g., DALY data points for other causes largely follow the diagonal), but NASH-related mortality data exhibits tail deviations, implying potential outlier influences (Figures S36-S41).

Figure 6 Historical trends and forecast of liver cancer burden by etiology in China, 1990–2036. The solid lines represent the observed historical trends in age-standardized rates from 1990 to 2021, while the dashed lines and shaded areas show the projected trends and their 95% uncertainty intervals for 2022–2036, generated using ARIMA models. ARIMA, Autoregressive Integrated Moving Average; AS(DALY)R, age-standardized disability-adjusted life year rate; ASIR, age-standardized incidence rate; ASMR, age-standardized mortality rate; NASH, non-alcoholic steatohepatitis.

Model goodness-of-fit metrics align with diagnostic plot findings: Ljung-Box test P values for all models (0.277–0.984) confirm no residual autocorrelation, supporting the appropriateness of model specifications. However, prediction interval diagnostics indicate systematic biases, with actual coverage rates (43.75–83.33%) significantly below the 95% target, correlating with the complex autocorrelation structures observed in ACF/PACF plots. Notably, alcohol-use-related incidence prediction performs poorest (43.75% coverage), directly linked to the multi-lag significant autocorrelation evident in its ACF plot (Tables S1,S2).

Projections of the BAPC model up to 2030 show that ASIR, ASMR, and AS(DALY)R for hepatitis HCV-associated liver cancer have continued to rise in both the male population and to decline in the female population; age-standardized rates of the various burden of disease metrics for the other disease types studied continue to decline for both males and females, except in the case of HCV-associated liver cancer, which has continued to decline in both males and females (Figures S42-S47).

Discussion

The study reveals a pivotal shift in the landscape of liver cancer in China from 1990 to 2021: the age-standardized rates of liver cancer (and liver cancer due to different etiologies)showed an overall decreasing trend, however, the burden of disease indicator showed a significant increase in absolute numbers, with population growth being the main factor for this change. Decomposition analysis showed that population aging contributed 141.91% to the increase in the absolute number of DALY for liver cancer between 1990 and 2021, and the population age structure contributed 54.7% of the positive driver, while the change in epidemiological trends offset 94.39% of the potential increase, revealing the importance of the change in the age structure of the population and aging in the prevention and control of liver cancer disease.

The observed decline in epidemiological rates likely reflects the monumental achievements in viral hepatitis control. The long-term prevalence of viral hepatitis is an important etiological basis for the high incidence of liver cancer in China, China has included the hepatitis B vaccine in the management of children’s programmed immunization since 1992, and it was included in the immunization program in 2002, which has reduced the hepatitis B surface antigen positive (HBsAg)-positive rate of children aged 1–4 years from 9.67% to 0.32% in 2014 through the strategy of universal newborn vaccination, and the HBsAg-positive rate of the adult population in China is 5–6% at present (26,27). Together with the iterative updating of antiviral drugs (e.g., entecavir, tenofovir) and the continuous optimization of therapeutic specifications, the incidence of HBV-related liver cancer has achieved a significant decline, fully confirming the effectiveness of the tertiary prevention system (27).

Simultaneously, breakthroughs in diagnosis and treatment have contributed to mitigating the burden.Breakthroughs in tumor diagnosis and treatment technologies have injected new momentum into disease control. In diagnostic imaging, the application of ultrasonography and hepatobiliary-specific MRI contrast has increased the detection rate of subclinical liver cancer to 63.5%; liquid biopsy technology has shifted the window for early diagnosis through circulating tumor DNA detection (27-31). In the therapeutic area, the popularity of precision hepatectomy has resulted in good survival benefits for patients with resectable liver cancer, with 5-year survival reaching 70.98%; and molecularly targeted agents (e.g., lenvatinib, regorafenib) and immune checkpoint inhibitors [programmed death-1 (PD-1)/programmed death-ligand 1 (PD-L1) monoclonal antibody] have significantly prolonged the survival of patients with inoperable resectable liver cancer (32-36). For example, sorafenib prolonged patients’ survival by 3 months compared to placebo, and by 2.3 months in the Asian population; it was the first targeted drug applied to liver cancer patients; then lenvatinib prolonged patients’ survival by 2 months on top of sorafenib (37,38). In recent years, the combination of targeted and immune drugs has also gradually become a new hot spot in the field of liver cancer research, and the combination regimen has extended the median survival of advanced patients to 19.2 months (39). These technological advances have pushed the age-standardized mortality rate of liver cancer in China down by 0.93% per year on average, and the DALY has decreased by 48.4%. However, the reconfiguration of the disease burden brought about by the demographic transition needs to be highly vigilant. The decline in HBV-related liver cancer underscores the success of primary prevention. For China’s vast chronic HBV population, implementing guideline-concordant surveillance is the next critical step. This entails lifelong, 6-monthly abdominal ultrasonography for high-risk groups (primarily those with cirrhosis or a family history of HCC), complemented by the strategic monitoring of alanine aminotransferase (ALT) and HBV DNA to guide antiviral therapy initiation—a key intervention for risk reduction. Strengthening this risk-stratified surveillance infrastructure is essential to further reduce mortality.

However, this study identifies a critical new challenge: metabolic factors are increasingly becoming a major contributor to the pathogenesis of liver cancer in China. The prevalence of NASH/metabolic dysfunction-associated steatotic liver disease (MASLD) in China has risen in parallel with the country’s rapid economic development, urbanization, and shifts toward Westernized diets and sedentary lifestyles. Recent large-scale studies estimate that the prevalence of NASH in China is approximately 29.2%, affecting over 300 million people, and it has become the most common chronic liver disease (40). There is a consensus that the burden of NASH-related HCC is increasing globally and in China (41).

These findings highlight that the prevention and control of liver cancer has entered a new stage of synergistic management of multiple etiological factors, which requires the establishment of a comprehensive prevention and control system covering viral control, metabolic interventions, and environmental management.

Limitations and future directions of this study

Our study relies on the modeled estimates from the GBD study, which may underestimate the cancer burden in ethnic minority areas or mobile populations. While GBD provides a powerful framework for comparisons across time and geography, it does not replace the need for detailed, individual-level cancer registry data. Registry-based studies, such as those conducted in other settings, can offer finer etiological resolution and better capture social inequalities (41-43). Future research utilizing China’s expanding cancer registry system is warranted to validate and refine our findings.

Another systematic concern is the relative insufficiency of high-quality, representative data from low-income regions and for ethnic minority populations within China. This uneven geographical and sociodemographic coverage in the underlying data (e.g., cancer registries, health surveys) poses a potential risk: the GBD models, which rely on available data and smoothing techniques, may underestimate the true incidence and mortality rates of liver cancer in these underserved populations. Consequently, our analysis might not fully capture the starkest health disparities, potentially leading to an over-optimistic view of the national situation and limiting the generalizability of our subnational and ethnicity-specific inferences. While the GBD study employs advanced statistical methods to correct for data gaps, this inherent limitation necessitates caution when applying our findings to resource-limited or minority settings.

Another significant limitation of this study stems from the inherent structure of the GBD data itself. The GBD reports HCC, ICC, and other pathological subtypes collectively as a single entity termed “liver cancer”. However, HCC and ICC exhibit fundamental differences in etiology, risk factors, population characteristics, and clinical prognosis. This consolidation may lead to several issues: first, it obscures the unique impact of distinct risk factors (e.g., HBV versus liver fluke infection) on specific subtypes, compromising the precision of causation attribution analyses; Second, the overall epidemiological burden and trends of liver cancer are primarily driven by the predominant HCC, potentially obscuring the unique geographic or temporal distribution patterns of ICC. Finally, given the starkly different prevention strategies for HCC and ICC, conclusions drawn from pooled data are difficult to translate into precise public health actions targeting specific subtypes. Future research requires registry data capable of distinguishing pathological subtypes to more deeply reveal the heterogeneity of liver cancer burden in China.

Furthermore, the ARIMA models inadequately capture time-series volatility patterns. Although baseline models pass residual independence tests, the insufficient prediction interval coverage underscores flaws in uncertainty quantification. Future studies should integrate more sophisticated variance structures (e.g., GARCH models) and customize model architectures based on etiology-specific autocorrelation traits to enhance predictive reliability. The BAPC model suffers from an identification problem due to collinearity among age, period, and cohort effects, requiring subjective constraints. The model assumes uniform effects across age groups, which may not reflect disease-specific patterns. It also inadequately captures sudden changes from events like policy interventions due to oversmoothing by its priors. Additionally, national-level data masks subnational variations, and standard error structures may not handle sparse data well. These issues necessitate cautious interpretation and complementary modeling approaches.

Conclusions

Although significant results have been achieved in the prevention and control of liver cancer in China (as evidenced by the decline in age-standardized burden), the accelerated aging of the population is becoming the primary reason for the continued increase in the absolute disease burden of liver cancer, while other factors other than HBV infection are becoming more and more important among the causative etiologies of liver cancer. Future precision prevention and treatment of liver cancer needs to take more account of the health management of the elderly population and the changing characteristics of the etiology of liver cancer, and efforts should be made to improve the completeness and representativeness of the disease surveillance data (especially in neglected populations), to assess and respond to the burden of liver cancer more comprehensively.

Acknowledgments

We appreciate the works of the Global Burden of Disease Study 2021 collaborators.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://cco.amegroups.com/article/view/10.21037/cco-25-80/rc

Peer Review File: Available at https://cco.amegroups.com/article/view/10.21037/cco-25-80/prf

Funding: This work was supported by the National Natural Science Foundation of China (No. 82260555 to W.Z.), and Major Science and Technology Projects of Gansu Province (No. 22ZD6FA021-4 to W.Z.).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://cco.amegroups.com/article/view/10.21037/cco-25-80/coif). W.Z. reports funding support from the National Natural Science Foundation of China (No. 82260555), and Major Science and Technology Projects of Gansu Province (No. 22ZD6FA021-4). The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
2. Ferlay J, Ervik M, Lam F, et al. Global Cancer Observatory: Cancer Today (Version 1.0). Int Agency Res Cancer 2024. Accessed February 1, 2024.
3. Ju W, Zheng R, Zhang S, et al. Cancer statistics in Chinese older people, 2022: current burden, time trends, and comparisons with the US, Japan, and the Republic of Korea. Sci China Life Sci 2023;66:1079-91. [Crossref] [PubMed]
4. Zheng R, Qu C, Zhang S, et al. Liver cancer incidence and mortality in China: Temporal trends and projections to 2030. Chin J Cancer Res 2018;30:571-9. [Crossref] [PubMed]
5. Cao W, Qin K, Li F, et al. Comparative study of cancer profiles between 2020 and 2022 using global cancer statistics (GLOBOCAN). J Natl Cancer Cent 2024;4:128-34. [Crossref] [PubMed]
6. Chen JG, Zhu J, Zhang YH, et al. Liver cancer mortality over six decades in an epidemic area: what we have learned. PeerJ 2021;9:e10600. [Crossref] [PubMed]
7. Chinese Society of Infectious Diseases, Chinese Medical Association. Chinese Society of Hepatology, Chinese Medical Association. Guidelines for the prevention and treatment of chronic hepatitis B (version 2019). J Clin Hepatol 2019;35:2648-69.
8. Xia C, Basu P, Kramer BS, et al. Cancer screening in China: a steep road from evidence to implementation. Lancet Public Health 2023;8:e996-e1005. [Crossref] [PubMed]
9. Lin J, Zhang H, Yu H, et al. Epidemiological Characteristics of Primary Liver Cancer in Mainland China From 2003 to 2020: A Representative Multicenter Study. Front Oncol 2022;12:906778. [Crossref] [PubMed]
10. Cao M, Ding C, Xia C, et al. Attributable deaths of liver cancer in China. Chin J Cancer Res 2021;33:480-9. [Crossref] [PubMed]
11. Devarbhavi H, Asrani SK, Arab JP, et al. Global burden of liver disease: 2023 update. J Hepatol 2023;79:516-37. [Crossref] [PubMed]
12. Ou TY, Huy LD, Mayne J, et al. Global mortality of chronic liver diseases attributable to Hepatitis B virus and Hepatitis C virus infections from 1990 to 2019 and projections to 2030. J Infect Public Health 2024;17:102443. [Crossref] [PubMed]
13. Wang H, Men P, Xiao Y, et al. Hepatitis B infection in the general population of China: a systematic review and meta-analysis. BMC Infect Dis 2019;19:811. [Crossref] [PubMed]
14. Dietz C, Maasoumy B. Direct-Acting Antiviral Agents for Hepatitis C Virus Infection-From Drug Discovery to Successful Implementation in Clinical Practice. Viruses 2022;14:1325. [Crossref] [PubMed]
15. Yu S, Wang H, Hu T, et al. Disease burden of liver cancer attributable to specific etiologies in China from 1990 to 2019: An age-period-cohort analysis. Sci Prog 2021;104:368504211018081. [Crossref] [PubMed]
16. Sohn W, Lee HW, Lee S, et al. Obesity and the risk of primary liver cancer: A systematic review and meta-analysis. Clin Mol Hepatol 2021;27:157-74. [Crossref] [PubMed]
17. Julien J, Ayer T, Bethea ED, et al. Projected prevalence and mortality associated with alcohol-related liver disease in the USA, 2019-40: a modelling study. Lancet Public Health 2020;5:e316-23. [Crossref] [PubMed]
18. Fang J, Celton-Morizur S, Desdouets C. NAFLD-Related HCC: Focus on the Latest Relevant Preclinical Models. Cancers (Basel) 2023;15:3723. [Crossref] [PubMed]
19. Šafčák D, Dražilová S, Gazda J, et al. Alcoholic Liver Disease-Related Hepatocellular Carcinoma: Characteristics and Comparison to General Slovak Hepatocellular Cancer Population. Curr Oncol 2023;30:3557-70. [Crossref] [PubMed]
20. Wang FZ, Zhang GM, Shen LP, et al. Comparative analysis of seroepidemiological survey results of hepatitis B among population aged 1–29 years in different endemic areas of China in 1992 and 2014. China J Prevent Med 2017;51:462-8. [PubMed]
21. Ross RK, Yuan JM, Yu MC, et al. Urinary aflatoxin biomarkers and risk of hepatocellular carcinoma. Lancet 1992;339:943-6. [Crossref] [PubMed]
22. Li Y, Ning Y, Shen B, et al. Temporal trends in prevalence and mortality for chronic kidney disease in China from 1990 to 2019: an analysis of the Global Burden of Disease Study 2019. Clin Kidney J 2023;16:312-21. [Crossref] [PubMed]
23. Rosenberg PS, Miranda-Filho A, Whiteman DC. Comparative age-period-cohort analysis. BMC Med Res Methodol 2023;23:238. [Crossref] [PubMed]
24. Lu M, Li D, Hu Y, et al. Persistence of severe global inequalities in the burden of Hypertension Heart Disease from 1990 to 2019: findings from the global burden of disease study 2019. BMC Public Health 2024;24:110. [Crossref] [PubMed]
25. Cai Y, Zhang J, Liang J, et al. The Burden of Rheumatoid Arthritis: Findings from the 2019 Global Burden of Diseases Study and Forecasts for 2030 by Bayesian Age-Period-Cohort Analysis. J Clin Med 2023;12:1291. [Crossref] [PubMed]
26. Luo Z, Li L, Ruan B. Impact of the implementation of a vaccination strategy on hepatitis B virus infections in China over a 20-year period. Int J Infect Dis 2012;16:e82-8. [Crossref] [PubMed]
27. Hou JL, Zhao W, Lee C, et al. Outcomes of Long-term Treatment of Chronic HBV Infection With Entecavir or Other Agents From a Randomized Trial in 24 Countries. Clin Gastroenterol Hepatol 2020;18:457-467.e21. [Crossref] [PubMed]
28. Calderaro J, Žigutytė L, Truhn D, et al. Artificial intelligence in liver cancer—new tools for research and patient management. Nat Rev Gastroenterol Hepatol 2024;21:585-99. [Crossref] [PubMed]
29. Shaik MR, Sagar PR, Shaik NA, et al. Liquid Biopsy in Hepatocellular Carcinoma: The Significance of Circulating Tumor Cells in Diagnosis, Prognosis, and Treatment Monitoring. Int J Mol Sci 2023;24:10644. [Crossref] [PubMed]
30. Park J, Lee YT, Agopian VG, et al. Liquid biopsy in hepatocellular carcinoma: Challenges, advances, and clinical implications. Clin Mol Hepatol 2025;31:S255-S284. [Crossref] [PubMed]
31. Bo Z, Song J, He Q, et al. Application of artificial intelligence radiomics in the diagnosis, treatment, and prognosis of liver cancer. Comput Biol Med 2024;173:108337. [Crossref] [PubMed]
32. Singal AG, Kanwal F, Llovet JM. Global trends in liver cancer epidemiology: implications for screening, prevention and therapy. Nat Rev Clin Oncol 2023;20:864-84. [Crossref] [PubMed]
33. Llovet JM, Pinyol R, Yarchoan M, et al. Adjuvant and neoadjuvant immunotherapies in liver cancer. Nat Rev Clin Oncol 2024;21:294-311. [Crossref] [PubMed]
34. Vitale A, Burra P, Frigo AC, et al. Survival benefit of liver resection for patients with liver cancer across different Barcelona Clinic Liver Cancer stages: a multicentre study. J Hepatol 2015;62:617-24. [Crossref] [PubMed]
35. Ming Y, Gong Y, Fu X, et al. Small-molecule-based targeted therapy in liver cancer. Mol Ther 2024;32:3260-87. [Crossref] [PubMed]
36. Vitale A, Romano P, Cillo U, et al. Liver resection vs nonsurgical treatments for patients with early multinodular liver cancer. JAMA Surg 2024;159:881-9. [Crossref] [PubMed]
37. Cheng AL, Kang YK, Chen Z, et al. Efficacy and safety of sorafenib in patients in the Asia-Pacific region with advanced liver cancer: a phase III randomised, double-blind, placebo-controlled trial. Lancet Oncol 2009;10:25-34. [Crossref] [PubMed]
38. Llovet JM, Ricci S, Mazzaferro V, et al. Sorafenib in advanced liver cancer. N Engl J Med 2008;359:378-90. [Crossref] [PubMed]
39. Galle PR, Finn RS, Qin S, et al. Patient-reported outcomes with atezolizumab plus bevacizumab versus sorafenib in patients with unresectable liver cancer (IMbrave150): an open-label, randomised, phase 3 trial. Lancet Oncol 2021;22:991-1001. [Crossref] [PubMed]
40. Liu J, Ayada I, Zhang X, et al. Estimating Global Prevalence of Metabolic Dysfunction-Associated Fatty Liver Disease in Overweight or Obese Adults. Clin Gastroenterol Hepatol 2022;20:e573-82. [Crossref] [PubMed]
41. Huang DQ, Singal AG, Kono Y, et al. Changing global epidemiology of liver cancer from 2010 to 2019: NASH is the fastest growing cause of liver cancer. Cell Metab 2022;34:969-977.e2. [Crossref] [PubMed]
42. Pinheiro PS, Jones PD, Medina H, et al. Incidence of Etiology-specific Hepatocellular Carcinoma: Diverging Trends and Significant Heterogeneity by Race and Ethnicity. Clin Gastroenterol Hepatol 2024;22:562-571.e8. [Crossref] [PubMed]
43. Pinheiro PS, Zhang J, Setiawan VW, et al. Liver Cancer Etiology in Asian Subgroups and American Indian, Black, Latino, and White Populations. JAMA Netw Open 2025;8:e252208. [Crossref] [PubMed]