Data on the burden of liver cancer attributable to HFPG in China from 1990 to 2021 were obtained from the GBD 2021, conducted by the Institute for Health Metrics and Evaluation (IHME) at the University of Washington. The GBD study provides comprehensive and comparable estimates of disease burden for 371 diseases and injuries, 88 risk factors, and 204 countries and territories (19, 20). To ensure data quality, all data sources underwent rigorous standardization and validation processes. For this study, we extracted data specific to China on liver cancer mortality, DALYs, YLDs, and YLLs attributable to HFPG. These data were stratified by sex, age group (in 5-year intervals from 25 to 29 to 95+), and year. Both all-age counts and age-standardized rates per 100,000 population were analyzed. HFPG-related burden estimates were derived from the comparative risk assessment (CRA) framework used in the GBD, which quantifies the disease burden attributable to specific risk factors by comparing observed health outcomes to those that would be expected under a theoretical minimum risk exposure level (19, 21). All estimates included 95% uncertainty intervals (UIs), calculated from the 2.5th and 97.5th percentiles of 1,000 draws from the posterior distribution, reflecting uncertainty from model inputs, sampling error, and measurement variability.

In the GBD 2021, liver cancer is defined as malignant neoplasms of the liver, encompassing International Classification of Diseases, 10th Revision (ICD-10) codes C22.0-C22.8 and a proportion of C22.9 consistent with primary liver cancer. This definition primarily includes hepatocellular carcinoma (HCC), the most common histologic subtype globally. HFPG is defined as a fasting plasma glucose concentration exceeding the theoretical minimum risk exposure level (TMREL) of 4.9–5.3 mmol/L (88.2–95.4 mg/dL), based on meta-analytical evidence of minimal risk for chronic disease outcomes and all-cause mortality at this range (19, 20). The estimation of liver cancer burden attributable to HFPG was based on the CRA framework developed by the IHME. This method involved calculating the population-attributable fraction (PAF) using the distribution of HFPG exposure in the population and the relative risks of liver cancer across exposure levels, compared against the TMREL. The PAFs were then applied to liver cancer mortality and disability estimates to derive the HFPG-attributable burden. Estimates included the number and age-standardized rates (per 100,000 population) of deaths, DALYs, YLDs, and YLLs, disaggregated by age, sex, and calendar year. DALYs were calculated as the sum of YLLs and YLDs.

The burden indicators, including DALYs, YLDs, and YLLs, along with the PAFs for liver cancer related to HFPG were calculated following the established GBD 2021 methodology (19, 20). Cause-specific mortality and YLLs were estimated using vital registration systems, cancer registries, and verbal autopsy data. YLDs were derived by multiplying the prevalence of each sequela (defined by ICD-10 codes C22.0-C22.8 and part of C22.9) by corresponding disability weights, estimated via population surveys. To model disease incidence, prevalence, and remission, the Bayesian meta-regression tool DisMod-MR 2.1 was employed. This approach uses a compartmental framework to ensure internal consistency among epidemiological parameters. Exposure levels for HFPG were estimated using a combination of large-scale national surveys and published datasets, synthesized with spatiotemporal Gaussian process regression (ST-GPR) models to generate continuous exposure surfaces by age, sex, location, and year. Theoretical minimum-risk exposure levels (TMREL, 4.9–5.3 mmol/L) and relative risks were obtained from meta-analyses of cohort studies, and these were used to calculate PAFs via comparative risk assessment (22). ASR (per 100,000 people) were calculated using the following formula: ASR = ∑ i = 1 N ai ∗ ωi ∑ i = 1 N ωi × 100 , 000 , where ai represents the age-specific rate in the i th age group, and ωi denotes the number (or weight) of individuals in the same age group in a standard population. DALYs were then computed as the sum of YLLs and YLDs attributable to HFPG. All estimates incorporated 95% UIs (2.5th-97.5th percentiles) derived from 1,000 model draws to reflect uncertainty in data inputs and model parameters (22).

A descriptive analysis was performed to assess the temporal and demographic patterns of liver cancer burden attributable to HFPG in China from 1990 to 2021. The analysis included estimates of both the all-age number and age-standardized rates (per 100,000 population) of deaths, DALYs, YLDs, and YLLs, stratified by sex and age group. Age-standardized rates were computed using the GBD global standard population to allow for consistent comparisons over time and between subgroups. Burden estimates for the year 2021 were first examined to describe the current distribution across age and sex. Age-specific counts and rates of deaths, DALYs, YLDs, and YLLs were plotted to identify peak burden ages and to assess sex disparities. To visualize long-term trends, annual estimates from 1990 to 2021 were graphed to evaluate changes over time in both absolute numbers and age-standardized rates. In addition, comparative analyses were conducted to place China’s trends within the global context. Age-standardized rates of HFPG-related liver cancer burden in China were compared to corresponding global values across the study period. These comparisons provided insights into how China’s epidemiological patterns align with or diverge from global trajectories.

Joinpoint regression analysis was employed to assess temporal trends in age-standardized mortality, DALYs, YLDs, and YLLs rates due to liver cancer attributable to HFPG in China from 1990 to 2021. This method identifies statistically significant changes in trend (i.e., joinpoints) and calculates the annual percent change (APC) for each segment, as well as the average annual percent change (AAPC) for the overall period. Analyses were conducted for both sexes combined and stratified by sex to explore gender-specific trend differences. The Joinpoint Regression Program (version 5.2.0) developed by the US National Cancer Institute was used for modeling. The number of joinpoints was selected using the Monte Carlo permutation method, with a maximum of five joinpoints allowed, and a significance level of 0.05 (23–25). Log-linear models were fitted to the age-standardized rates to ensure constant variance, and APCs were considered statistically significant when the 95% confidence interval did not include zero. This approach enabled the detection of non-linear temporal patterns and the quantification of time periods with accelerating or decelerating trends in HFPG-attributable liver cancer burden. The use of joinpoint regression complements descriptive analyses by offering more precise insights into how the burden has evolved in response to demographic, epidemiological, and health system changes over time.

To disentangle the temporal effects of age, calendar period, and birth cohort on the trends in liver cancer burden attributable to HFPG in China, an APC analysis was conducted. This approach allows for the assessment of how risk varies across different age groups, time periods, and generational cohorts while controlling for the linear dependency among these factors (26, 27). APC models were applied to age-standardized mortality and DALY rates stratified by sex, using data from 1990 to 2021. Age was grouped into 15 five-year intervals from 25–29 to 95+, and calendar periods were divided into 5-year intervals from 1992–1996 to 2017–2021. Birth cohorts were calculated by subtracting the midpoint of each age group from the midpoint of each calendar period. The intrinsic estimator (IE) method was used to address the identification problem arising from the exact linear relationship among age, period, and cohort. This method has been widely adopted in epidemiological research for its ability to generate unbiased and interpretable coefficients. The APC model fitting was performed using the “Epi” package (version 2.51) in R (version 4.3.1), and the residual deviations between models and Akaike Information Criterion (AIC) were compared to determine the optimal model.

To quantify the contributions of demographic and epidemiological factors to changes in the absolute burden of liver cancer attributable to HFPG in China from 1990 to 2021, a decomposition analysis was conducted. This method partitions the total change in the number of deaths and DALYs into three components: population growth, population aging, and epidemiological changes (28). First, the difference in absolute numbers of HFPG-attributable deaths and DALYs between 1990 and 2021 was calculated. Using the Das Gupta method, the contribution of each individual factor was estimated while keeping the other two variables constant. Population growth reflects changes in total population size, aging captures shifts in the population age structure, and epidemiological change accounts for differences in age-specific and sex-specific HFPG-attributable liver cancer rates. These epidemiological changes reflect shifts in disease burden driven by evolving exposures to metabolic risk factors, improvements or deteriorations in public health infrastructure, changes in disease screening and diagnosis, and advancements in treatment strategies. This analysis was performed for the total population and stratified by sex to evaluate gender-specific drivers of change. The decomposition results provide insights into the relative influence of demographic dynamics versus temporal changes in population health and healthcare delivery. All decomposition calculations were performed using R software (version 4.3.1), and estimates were visually presented using bar plots to highlight the proportional contributions of each component over time.

In 2021, the burden of liver cancer attributable to HFPG in China remained substantial, with a notable sex disparity. A total of 3,467 deaths were estimated to be linked to HFPG, with a higher burden observed in males than females. This trend was consistently reflected across other health loss metrics. The total DALYs attributable to HFPG reached 83,113, with males accounting for a greater proportion of this burden than females. YLLs made up the vast majority of DALYs, underscoring the fatal nature of HFPG-related liver cancer, whereas YLDs contributed minimally. When standardized for age, the burden remained higher in men than in women, with age-standardized DALY and YLL rates per 100,000 population in males approaching 4.5 and 4.46, respectively, compared to 3.25 and 3.21 in females (Table 1).

In 2021, the burden of liver cancer attributable to HFPG in China demonstrated pronounced variations by age and sex. The number of deaths rose sharply in older age groups, peaking at 65–69 years for both sexes, with males showing higher death counts across all age groups. DALYs followed a similar trend, increasing significantly with age and reaching their highest levels in the same age group, again with males bearing a greater burden. YLDs peaked at ages 65–69 in both sexes, while YLLs peaked at 55–59 and 65–69 years in males and at 65–69 and 70–74 years in females, underscoring the impact of premature mortality in these groups. When age-standardized rates were examined, all indicators (deaths, DALYs, YLDs, and YLLs) exhibited steep increases with age, particularly peaking around 85–89 years for both sexes. Across all measures, males consistently experienced higher rates than females, indicating a heavier burden of HFPG-related liver cancer among men (Figure 1 and Supplementary Figure S1).

Between 1990 and 2021, the number of deaths, DALYs, YLDs, and YLLs attributable to HFPG in China shows an overall fluctuating increase from 1990 to 2021, with observable variations at different time points. This upward trend in absolute numbers is more pronounced among males, who consistently exhibit a higher burden across all indicators. In contrast, the age-standardized rates of deaths, DALYs, YLDs, and YLLs have remained relatively stable throughout the period, suggesting that population growth and aging have primarily driven the increase in total burden. Notably, a slight downward trend in age-standardized rates has emerged in recent years across all four measures, indicating potential improvements in risk factor management or early intervention (Figure 2 and Supplementary Figures S2A,B).

From 1990 to 2021, older populations have consistently experienced a heavier burden of liver cancer attributable to HFPG in China, with the increase most prominent among individuals aged 75 years and above. As illustrated in Figure 2 and Supplementary Figure S2, not only have mortality rates risen substantially in these age groups, but YLDs and overall disease impact have also expanded, underscoring a growing burden in the aging demographic. While HFPG-related liver cancer remains present across all age groups, the trend indicates a shifting concentration of burden toward older adults. This transition is particularly evident in the sharp rise in crude death rates and YLLs among the elderly, reflecting an increasing role of premature mortality in driving the disease burden (Figure 2 and Supplementary Figures S2C,D).

Between 1990 and 2021, age-standardized rates of liver cancer attributable to HFPG remained relatively stable in China but increased significantly at the global level. As shown in Table 2 and Supplementary Figure S3, the ASMR, DALYs, YLDs, and YLLs rates in China showed only marginal changes over the three decades, with minor fluctuations and no statistically significant trends. For instance, the ASMR in China increased slightly from 0.16 to 0.17 per 100,000, while DALYs and YLLs rates decreased marginally. In contrast, global rates for all four indicators rose significantly, with statistically meaningful increases, including a 59% increase in ASMR and a 1.81-unit rise in age-standardized DALY rates. The most prominent global increase was seen in YLDs, which more than doubled. These findings suggest that although the absolute burden in China has grown, the age-specific risk has remained relatively steady, unlike the global context where HFPG-related liver cancer risk has escalated across populations.

Joinpoint regression analysis revealed distinct temporal patterns in age-standardized rates of liver cancer attributable to HFPG in China from 1990 to 2021, with notable differences by sex and indicator type. As shown in Figure 3, Supplementary Figure S4 and Supplementary Table S1, the trends were characterized by multiple phases of significant increases and decreases. Among both sexes, a sharp rise in all indicators occurred between 1995 and 2000, followed by a significant decline during the early 2000s, and a subsequent increase from the mid-2000s to mid-2010s. In recent years, a downward trend has emerged, with statistically significant declines observed from 2017 onward in DALYs and YLLs, particularly in females (AAPC = −7.82%) and to a lesser extent in males. Notably, while age-standardized YLD rates exhibited modest increases over the entire period (AAPC = 0.77% for both sexes), mortality and YLL rates remained largely stable in the long term, with average annual percent changes close to zero.

Age-specific patterns show that both mortality and DALYs increase with advancing age, indicating a heavier burden in older populations. Period-specific analyses reveal variations across different time periods, reflecting the influence of societal changes, healthcare advancements, and shifts in HFPG prevalence over time. The cohort effects underscore that more recent birth cohorts experience higher rates of mortality and DALYs as they age compared to earlier cohorts, suggesting a rising vulnerability in younger generations as they reach older age. The analysis of birth cohort-specific trends across different periods further emphasizes that individuals born in more recent cohorts face a greater burden as they age, highlighting the long-term impact of changing metabolic risks (Figure 4 and Supplementary Figure S5).

As shown in Figure 5, decomposition analysis revealed distinct patterns in the driving forces behind changes in HFPG-related liver cancer burden in China from 1990 to 2021, with notable gender differences. For deaths, aging emerged as the most significant contributor, followed by epidemiological changes and population growth, with all three factors exerting a more pronounced influence among men. In contrast, the decomposition of DALYs showed that population growth was the leading driver, followed by aging and then epidemiological changes. Interestingly, while all three components acted as positive drivers among men, epidemiological change served as a protective factor for women, mitigating their overall DALY burden.