Introduction
Liver cancer is the sixth most common cancer and the third leading cause of cancer-related deaths worldwide, resulting in 905,700 new cases and 830,200 deaths in 2020.[1] Globally, the incidence and mortality rates of liver cancer have declined dramatically in the past three decades.[2] However, large geographic disparities have been found in trends in liver cancer rates. The incidence and mortality rates of liver cancer have been rising in several formerly low-rate areas, such as the USA, Australia, and parts of Europe, but decreasing in formerly high-burden countries, most located in Asia and Africa.[3,4] This huge heterogeneity is mainly determined by the geographical variation in the trends of liver cancer risk factors.[5-7] The applications of hepatitis B virus (HBV) vaccination and hepatitis C virus (HCV) antiviral therapy have directly led to a rapid decline in the burden of liver cancer in the Asia-Pacific region in recent years,[8,9] while the increasing trends in the American and some European countries are mainly attributed to the rise of alcohol abuse, HCV infection, obesity, and nonalcoholic fatty liver disease.[10,11] A comprehensive understanding of variation in temporal patterns of liver cancer burden and underlying etiologies will greatly aid in developing tailored intervention strategies.
Previous studies have mainly explored the homogeneity and heterogeneity in the liver cancer burden trends across predefined socioeconomic and geographic subgroups,[12,13] which may have missed marked discrepancies. To the best of our knowledge, few studies have used a data-driven method to describe disparities in the changes in the liver cancer burden across countries. Therefore, in this study, we fill this research gap by characterizing the 30-year trajectories of liver cancer burden across countries using growth mixture models (GMMs). This method classifies countries or territories into subgroups that share a common trajectory of liver cancer burden over time. Moreover, we explored the key risk factors that drive changes in the liver cancer burden over the past three decades and the potential socioeconomic determinants. Furthermore, we predicted the future liver cancer burden through 2035.
Methods
Data sources
Data on annual new cases, number of deaths, incidence rate, and mortality rate of liver cancer in 204 countries and territories from 1990 to 2019 by year, sex, location, age (ages 0–84 years in 5-year increments and 85 years and older), and etiology were retrieved using the Global Health Data Exchange query tool (https://vizhub.healthdata.org/gbd-results/). The general methods of the Global Burden of Disease (GBD) study and the methods for estimating the liver cancer burden have been detailed in previous studies.[2] Briefly, liver cancer incidence and mortality in the GBD dataset were estimated as follows. (1) The mortality-to-incidence ratio (MIR) for liver cancer was estimated using matched incidence and mortality data and the healthcare access and quality index. (2) A Cause-of-Death Ensemble model was used to generate liver cancer mortality estimates using data of vital registration systems, cancer registries, verbal autopsy reports, and related covariates. (3) Liver cancer mortality estimates were scaled to align with total mortality from all causes of death. (4) The adjusted mortality estimates were divided by the MIR estimates from the first step to obtain incidence estimates. Moreover, to split the liver cancer burden into the five major etiology groups included in GBD 2019 (hepatitis B, hepatitis C, alcohol use, nonalcoholic steatohepatitis, and other causes), a systematic literature search was performed in PubMed to obtain representative proportion data. The proportions and relevant risk factor data were then combined using the DisMod-MR model to estimate the final proportion of the data. Finally, the proportion of the data was rescaled to a total of 100%.
As social and economic determinants, we selected gross domestic product (GDP) per capita and current health expenditure per capita in purchasing power parity international dollars (CHE per capita in Purchasing Power Parities current international $ [PPP int. $]) and universal health coverage (UHC) service coverage index as macro indicators of socioeconomic status. These three socioeconomic status indicators were chosen as proxy measures of the levels of economic development, health investment, and healthcare access.[14,15] Moreover, the Socio-demographic Index (SDI), a summary indicator of income per capita, years of schooling, and fertility rate in females younger than 25 years, was also used.
As these socioeconomic indicators may change over time, the latest available data with the lowest missing rates were chosen for comparison. Data on the SDI in 2019 were extracted from the GBD study database,[16] CHE per capita in PPP int. $, the UHC service coverage index in 2019 from the WHO Global Health Observatory Data Repository,[17] and the GDP per capita in 2019 from the World Bank DataBank.[18]
The GBD world population standard was used to calculate age-standardized rates. To predict the liver cancer burden, we used the latest population estimates provided by the United Nations Economic and Social Council for 2022, which were available for all countries and territories included in our study (https://population.un.org/wpp/Download/Standard/Population/).
Statistical analysis
The trajectories of the age-standardized incidence rate (ASIR) and age-standardized mortality rate (ASMR) of liver cancer over the last 30 years were defined using GMMs with random intercepts. We fitted the GMMs to the transformed values of age-standardized rates using the Box-Cox transformation. I-splines with three equidistant knots were used as link functions in all models. To determine the optimal number of latent trajectories, we started with a one-class GMM and successively increased the number of classes in subsequent models until the best solution was identified or the number of classes reached five. Model selection focused on fit indices (Bayesian information criterion [BIC] and Akaike information criterion [AIC]) and values of mean posterior class membership probabilities, and substantive interpretability. The fit indices, average predicted trajectories, and percentages of each latent trajectory for the competing GMMs are presented in Supplementary Figures 1 and 2. Supplementary Table 1, https://links.lww.com/CM9/B561 shows the number of parameters and mean posterior class membership probabilities in the competing GMMs. If the improvement in fit statistics with the addition of a latent trajectory was relatively small, the most parsimonious models were selected. Models that included classes with approximately 5% of the sample were not considered because of their lack of robustness.[19]
Sex-specific trajectory modeling was also performed for the liver cancer ASIR and ASMR (Supplementary Figures 3–6 and Supplementary Tables 2 and 3, https://links.lww.com/CM9/B561), and the results were similar to those of the analyses of the total population (Supplementary Figures 7–8, http://links.lww.com/CM9/B561). Therefore, to increase statistical power, analyses of different sexes are only displayed in the supplementary materials, http://links.lww.com/CM9/B561. F.Y. and D.S. independently screened these fit indices, percentages of each latent class, and mean posterior class membership probabilities and decided on the final solutions for all GMMs. Any disagreements in the model selection process were resolved by a third investigator (W.C.).
To investigate the driving forces underlying liver cancer trajectories at the level of risk factors and socioeconomic determinants, we first displayed the temporal trends of ASIR and ASMR according to five major risk factors in different trajectory groups (hepatitis B, hepatitis C, alcohol use, nonalcoholic steatohepatitis, and other causes) and calculated the contributions of different risk factors to the change in ASIR or ASMR from 1990 to 2019. This indicator is defined as , where denotes the total liver cancer ASIR or ASMR in 2019 and denotes the liver cancer ASIR or ASMR due to specific risk factors in 2019. We then drew boxplots of SDI, GDP per capita, CHE per capita, and the UHC service coverage index according to the identified latent trajectories. The Wilcoxon rank-sum test with Bonferroni adjustment for multiple testing was conducted to examine the associations between trajectory membership and these socioeconomic variables. All tests were two-tailed, and statistical significance was set at P <0.05.
A Bayesian age-period-cohort model with integrated nested Laplace approximation was used to predict the age-standardized liver cancer rate and the number of new cases and deaths at national and regional levels through 2035.[20] This model has shown better accuracy and precision than other prediction methods.[21,22] It decomposes the observed liver cancer rate multiplicatively into the general level, age, period, and cohort effects and applies Gaussian autoregressive priors to project the posterior distribution of period and cohort effects. The first-order random walk (RW1) prior assumes a constant trend over the time scale, whereas the second-order random walk 2 (RW2) prior assumes a linear time trend. As models with RW2 priors outperformed those with RW1 priors in previous studies[23] and were thus selected for our projection. We performed all statistical analyses using R software (version 4.0.1 R Foundation for Statistical Computing, Vienna, Austria). GMMs and Bayesian age-period-cohort model were conducted using packages "lcmm"[24] and "BAPC".[20]
Results
Globally, the absolute number of new liver cancer cases has increased from 373,400 in 1990 to 534,400 in 2019 (Table 1). The global ASIR displayed a decreasing trend, with an estimated annual percentage change (EAPC) of –1.93 (95% confidence interval [CI]: –2.31, –1.55). Large heterogeneity in ASIR trends across countries was observed, and three discrete trajectories were identified via comprehensive considerations of model fit and substantive interpretability. The fit statistics improved with the addition of a fourth class, but the models included a class with a sample size of approximately 5%, which was meaningless in the subsequent exploratory analysis. Hence, the three-class solution appears to be optimal for ASIR. Most countries and territories in our study (64.2%, n = 131) followed a trajectory in which the average predicted ASIR remained moderate between 1990 and 2019 (stable group) (Table 1 and Supplementary Figure 1, https://links.lww.com/CM9/B561). However, the total number of new cases in these countries has increased from 90,300 in 1990 to 197,400 in 2019. Most of the countries and territories in the Eastern Mediterranean (95.2%, n = 20), African (85.1%, n = 40), Southeast Asia (81.8%, n = 9), and Western Pacific Region (74.1%, n = 20) were classified in this class (Figure 1). The second trajectory, referred to as the decreasing group (14.7%, n = 30), consisted of countries and territories that reported a high liver cancer burden in the 1990s, followed by a rapid decrease (Supplementary Figure 1, https://links.lww.com/CM9/B561). However, the ASIR of these countries and territories remained the highest among the three classes during the study period (22.82 per 100,000 in 1990 and 9.57 per 100,000 in 2019), and the number of new cases accounted for 67.7% and 42.8% of global liver cancer cases in 1990 and 2019, respectively. Compared with other WHO regions, the percentage of this class in the American region was the highest (48.6%, n = 17) (Figure 1). Class 3 consisted of 43 (21.1%) countries and territories and was characterized by an increasing trend in the average predicted ASIR during the study period (increasing group) (Table 1 and Supplementary Figure 1, https://links.lww.com/CM9/B561). From 1990 to 2019, the ASIR of these countries and territories increased by an average of 3.22% (95% CI: 2.86–3.58%) per year (from 2.74 per 100,000 in 1990 to 6.10 per 100,000 in 2019), and the new cases increased by 254.9% from 30,500 in 1990 to 108,100 in 2019. Almost half of the countries and territories in the European region (49.1%, n = 26) belonged to this class (Figure 1).
Table 1 -
The new cases, deaths, ASIR, and ASMR of liver cancer according to trajectory groups in 1990 and 2019, and its temporal trends from 1990 to 2019.
|
Trajectory
group
*
|
N of countries (%)
|
1990
|
|
2019
|
|
1990–2019
|
|
N ×103
(95% UI)
|
ASR
per 100,000 (95% UI)
|
|
N ×103
(95% UI)
|
ASR
per 100,000 (95% UI)
|
|
EAPC
†
(95% CI)
|
|
Incidence
|
|
|
|
|
|
|
|
| Increasing |
43 (21.1) |
30.5 (24.9, 39.3) |
2.74 (2.22, 3.56) |
|
108.1 (79.6, 153.2) |
6.10 (4.46, 8.74) |
|
3.22 (2.86, 3.58) |
| Stable |
131 (64.2) |
90.3 (73.3, 117.2) |
4.94 (4.05, 6.33) |
|
197.4 (146.7, 277.5) |
5.00 (3.74, 6.99) |
|
–0.29 (–0.53, –0.06) |
| Decreasing |
30 (14.7) |
252.6 (206.6, 314.8) |
22.82 (18.73, 28.32) |
|
228.7 (185.2, 288.6) |
9.57 (7.76, 12.06) |
|
–4.48 (–5.25, –3.70) |
| Global |
204 (100) |
373.4 (335.9, 415.7) |
8.98 (8.10, 9.97) |
|
534.4 (486.5, 588.6) |
6.51 (5.95, 7.16) |
|
–1.93 (–2.31, –1.55) |
|
Death
|
|
|
|
|
|
|
|
|
| Increasing |
33 (16.2) |
23.0 (19.8, 27.7) |
2.24 (1.91, 2.71) |
|
76.7 (62.8, 96.4) |
4.66 (3.79, 5.90) |
|
2.93 (2.33, 3.31) |
| Stable |
137 (67.2) |
66.9 (49.5, 95.3) |
4.14 (3.10, 5.80) |
|
159.7 (116.0, 231.4) |
4.40 (3.23, 6.30) |
|
0.18 (0.09, 0.26) |
| Decreasing |
34 (16.7) |
275.2 (229.4, 336.6) |
20.36 (17.04, 24.81) |
|
248.0 (205.2, 306.2) |
8.80 (7.28, 10.88) |
|
–4.23 (–4.87, –3.58) |
| Global |
204 (100) |
365.2 (330.0, 405.8) |
8.93 (8.09, 9.90) |
|
484.6 (444.1, 525.8) |
5.95 (5.44, 6.44) |
|
–2.23 (–2.63, –1.83) |
* New cases, deaths, and ASR in each latent trajectory were generated using Monte Carlo simulation by assuming a log-normal distribution of age-specific rates in each country and territory.† EAPC was calculated according to a regression model fitted to the natural logarithm of the ASR, namely ln(ASR) = α + β × (calendar year)ε. EAPC was defined as 100 × (exp(β)-1) and its 95% confidence interval (95% CI) was calculated in the fitted model. ASIR: Age-standardized incidence rate; ASMR: Age-standardized mortality rate; ASR: Age-standardized rate; CI: Confidence interval; EAPC: Estimated annual percentage change; UI: Uncertainty interval.
;)
Figure 1: The distribution of liver cancer trajectories across World Health Organization regions for incidence and mortality.
A three-class GMM was chosen as the final model for liver cancer ASMR. The average predicted trajectories of ASMR had a pattern similar to that of ASIR (Supplementary Figure 2, https://links.lww.com/CM9/B561) but with more countries and territories in the stable or decreasing groups (Table 1). The largest of these trajectories was the stable group (67.2%, n = 137) with relatively stable ASMR (EAPC = 0.18, 95% CI: 0.09, 0.26). The increasing group (16.2%, n = 33) reported a low ASMR in the 1990s and experienced pronounced increases in the following decades. The decreasing group (16.7%, n = 34) was initially characterized by countries and territories with a heavy burden, followed by a decline from 20.36 per 100,000 in 1990 to 8.80 per 100,000 in 2019. As for the absolute number, more than half of liver cancer deaths in 1990 and 2019 occurred in countries and territories in the decreasing group. The geographical distribution of ASMR trajectories was similar to that of ASIR, with the increasing group being most common in the European region (37.7%, n = 20) and the decreasing group in the American region (48.6%, n = 17) (Figure 1).
HBV was the most important driver of the downward trend in the decreasing group, contributing to 63.4% of the change in the total liver cancer ASIR from 1990 to 2019 (Figure 2 and Supplementary Table 4, https://links.lww.com/CM9/B561). The ASIR of liver cancer due to HBV (LCHB) was the highest among the five etiology-specific ASIR in 2019 (5.90 per 100,000). The upward trend in ASIR in the increasing group was mainly driven by alcohol use, HCV infection, and HBV infection, and its respective etiology-specific ASIR increased by 1.03, 1.05, and 0.81 per 100,000 in the past three decades, which contributed to 30.8%, 31.1%, and 24.2% of the change in total liver cancer ASIR. The ASIR of liver cancer due to alcohol use, liver cancer due to HCV (LCHC), and LCHB also accounted for 85.6% of the total ASIR in the increasing group in 2019. The magnitude of change in etiology-specific ASIR varied in the stable group and was relatively small compared with other groups, with the most pronounced increase observed in the ASIR of liver cancer due to nonalcoholic steatohepatitis (0.08 per 100,000) and the most pronounced decrease observed in the ASIR of LCHB (–0.07 per 100,000). Nevertheless, the highest ASIR due to specific risk factors in this group in 2019 was that of LCHC (2.05 per 100,000). The ASMR patterns mirrored those of ASIR. HBV was the main driver in the decreasing group (60.4%) and still caused a high liver cancer ASMR in 2019 (4.55 per 100,000) (Figure 2 and Supplementary Table 5, https://links.lww.com/CM9/B561). The ASMR of LCAL increased faster than that of the other four risk factors in the increasing group, accounting for 33.7% of the change between 1990 and 2019. Alcohol consumption was also a major risk factor in this group in 2019 (1.50 per 100,000). While the ASMR of LCHC was the highest among the five etiology-specific ASMRs in the stable group in 2019 (1.48 per 100,000), the most pronounced change during the study period was observed in the ASMR of LCAL (0.13 per 100,000). At the country level, alcohol use was the most important driver of liver cancer burden changes in most of the increasing group countries, whereas HBV and alcohol use were the most important drivers in the majority of the decreasing group countries (Supplementary Table 6, https://links.lww.com/CM9/B561).
Figure 2: The ASIR (A) and ASMR (B) of liver cancer caused by five etiologies, according to trajectory groups, from 1990 to 2019. The bar at the top-right of each subgraph represents the contributions of different etiologies to the change in ASIR or ASMR from 1990 to 2019, which is not displayed if the contributions of different etiologies operate in different directions. ASIR: Age-standardized incidence rate; ASMR: Age-standardized mortality rate; LCAL: Liver cancer due to alcohol use; LCHB: Liver cancer due to hepatitis B; LCHC: Liver cancer due to hepatitis C; LCNS; Liver cancer due to non-alcoholic steatohepatitis; LCOT: Liver cancer due to other causes.
Figure 3 shows the socioeconomic patterns by the ASIR and ASMR trajectories. A substantial difference was observed between the increasing and stable groups. Countries with increasing incidence and mortality tended to have higher SDI, GDP per capita, CHE per capita, and UHC service coverage index (all P<0.05). Moreover, the socioeconomic situation of countries in the decreasing group was slightly better than that of countries in the stable group; however, the difference was not statistically significant for most indicators.
Figure 3: Distribution of socioeconomic indicators according to liver cancer trajectories for incidence (A) and mortality (B). Data for 19,17, and 11 countries or territories are removed in the subgraphs of GDP per capita, CHE per capita, and UHC due to data integrity. P values were based on Wilcoxon's rank sum test with Bonferroni adjustment (* P < 0.05, † P < 0.001). CHE per capita: Current health expenditure per capita; GDP per capita: The gross domestic product per capita; PPP int.: Purchasing Power Parities international; SDI: Socio-demographic Index; UHC: Universal health coverage. Each dot represents data of a country.
Figures 4A,B show the observed and predicted ASIR and ASMR by trajectory from 1990 to 2035, respectively. Globally, the ASIR will increase slightly (from 6.51 per 100,000 in 2019 to 6.66 per 100,000 in 2035), while the ASMR will decrease (from 5.95 per 100,000 to 5.69 per 100,000 in 2035). The historical downward trend in the decreasing group is reversed for ASIR (from 9.57 per 100,000 in 2019 to 10.93 per 100,000 in 2035) and ASMR (from 8.80 per 100,000 in 2019 to 8.91 per 100,000 in 2035) over the next decade. The trends in the ASIR and ASMR observed in the increasing group are predicted to ease, reaching 6.41 per 100,000 and 4.81 per 100,000, respectively, by 2035. A minor decreasing trend occurred in the stable group, and the ASIR and ASMR of this group will remain the lowest among the three classes in 2035 (4.80 per 100,000 and 4.02 per 100,000, respectively). Regarding the absolute number, increasing trends were detected in all three trajectories, with the most pronounced increase in new cases in the decreasing group (56.80%) and deaths in the stable group (49.66%) (Figure 4C,D). Supplementary Figures 9–12, https://links.lww.com/CM9/B561 display the top 15 countries with the highest number of new cases and deaths of liver cancer in 2019, accounting for 80.2% and 78.4% of the global total, respectively. These countries or territories are located in different regions and have experienced various changes in the patterns of liver cancer burden between 1990 and 2019. Most selected countries are predicted to decline or remain stable in ASIR and ASMR by 2035; however, a consistently increasing trend in new cases and deaths will be found across nearly all countries. The most pronounced increase in ASIR (EAPC = 1.14, 95% CI: 1.08–1.21) and the greatest increase in new cases (57.9%) would be observed in the USA. In contrast, the lowest EAPC for ASIR (EAPC = –1.78, 95% CI: –1.81 to –1.74) was detected in Egypt, and the greatest decrease in new cases (–1.78%) was observed in France. In terms of ASMR, the highest EAPC (EAPC = 1.24, 95% CI: 1.16–1.32) and the greatest increase in deaths (68.9%) would be detected in the USA. The most pronounced decrease in ASMR (EAPC = –2.05, 95% CI: –2.09 to –2.01) was observed in Egypt, whereas the least increase in deaths (2.99%) was observed in France.
Figure 4: The temporal trends of disease burden of liver cancer, by trajectory groups, between 1990 and 2019 and their projections to 2035. (A) ASIR. (B) ASMR. (C) Number of new cases. (D) Number of deaths. The vertical dashed line indicates where the prediction starts. ASIR: Age-standardized incidence rate; ASMR: Age-standardized mortality rate.
Discussion
Our study provides a comprehensive yet concise profile of the change patterns in the liver cancer burden. Three discrete trajectories of liver cancer ASIR and ASMR have been identified worldwide as increasing, stable, and decreasing groups. Most countries and territories followed stable ASIR and ASMR trajectories. HBV was the major driver in the decreasing group, whereas alcohol use, HCV, and HBV contributed the most to the upward trend in the increasing group. Countries and territories in the increasing group have overwhelming advantages in terms of socioeconomic status. Increasing trends in this group are expected to ease in the next decade. Unfortunately, the ASIR and ASMR of the decreasing group were predicted to remain at their current elevated levels.
Consistent with previous studies,[25,26] an increased trajectory of the incidence and mortality of liver cancer was found in many parts of the world in recent years, such as the USA, Australia, and many European countries, whereas decreasing trends were observed in some Asian countries, including China, the Philippines, Japan, and Vietnam (Figure 1 and Supplementary Figures 9–12, https://links.lww.com/CM9/B561). Geographical variations in risk factor trends may be responsible for this heterogeneous pattern. The etiology of liver cancer has been extensively investigated. Different infectious and noninfectious factors, including viral infections (HBV and HCV), behavioral factors (excess alcohol consumption and smoking), metabolic factors (obesity, type 2 diabetes, and nonalcoholic steatohepatitis), and aflatoxin exposure, are associated with the onset of liver cancer. Infectious factors were eminent causes of liver cancer globally but have diminished in recent years. Our study indicated that HBV was the major driving force underlying the patterns of change in the decreasing group, whereas the upward trend in the increasing group was mainly driven by alcohol use, HCV, and HBV infection. Previous evidence has also confirmed that in some parts of Asia, the combination of successful mass vaccination against HBV and reduction in aflatoxin exposure contributed to a sharp reduction in the prevalence of hepatitis B surface antigen and a concomitant decrease in liver cancer burden.[27] Moreover, the downward trend in the HBV-related liver cancer rate can be expected to accelerate in the next few decades as the vaccinated cohort ages. In contrast, recent studies have shown that the major driving force behind the upward trend in some affluent countries, including the USA and several European countries, is the high HCV infection prevalence due to the receipt of contaminated blood and injection drug use.[10,28,29] In addition, increased alcohol use[7] and an elevated prevalence of metabolic factors[30] in these regions may act synergistically with HCV to contribute to the rising burden of liver cancer,[31,32] and these factors also independently increase the risk of liver cancer. Furthermore, the Eastern European countries, Poland, and the Russian Federation demonstrated rising hepatitis B surface antigen prevalence, as did some high-income countries like Australia and Canada.[33] These stress the importance of prevention and treatment of HCV infection, behavioral risk factor management, and universal hepatitis B immunization in these countries.
Although compared with the increasing and decreasing groups, most countries and territories in the stable group had consistently lower ASIR and ASMR, and several countries in this group suffered from high liver cancer incidence and mortality rates during the study period, such as Thailand and Egypt (Table 1 and Supplementary Figures 9–12, https://links.lww.com/CM9/B561). The persistent prevalence of country-specific carcinogenic agents is likely to be responsible for these trends. For example, the trend of liver cancer in Egypt is mainly driven by the consistently high prevalence of HCV infections owing to historical iatrogenic transmission. Fortunately, efforts have been made to counteract this,[34] and the liver cancer burden is expected to continue to decline in the near future. In contrast, the key determinants of liver cancer in Thailand are the endemic bile duct flukes Opisthorcis viverrine and Clonorchis sinesis, which lead to intrahepatic cholangiocarcinoma, the second subtype of liver cancer behind hepatocellular carcinoma. Intensive health education, ecosystem monitoring, and treatment of liver fluke infections should be prioritized to prevent new intrahepatic cholangiocarcinoma cases and deaths.
Over the next decade, the number of new cases and deaths in the three groups will continue to rise owing to population aging and growth. The historical decreasing trend in ASIR and ASMR observed in the decreasing group will be reversed and still remain highest among the three trajectory groups. Given the increasing prevalence of alcohol use[7] and metabolic disorders[35] and recent declines in the prevalence of HBV and HCV in these regions,[36] it is reasonable to speculate that non-infection-related liver cancer will offset the gains achieved by infection control. In China, the rates of liver cancer have begun to plateau in recent birth cohorts,[37] partly due to the acquisition of sedentary lifestyles, alcohol use, and metabolic risk factors in parallel with the rapid socioeconomic transition. Manthey et al[38] estimated that alcohol use has increased substantially in several lower-middle- and upper-middle-income countries since 1990, such as China, India, and Vietnam, surpassing the use levels in some European countries in 2017, and this trend will continue until 2030. A recent global modeling analysis showed that children in high-income countries had the lowest prevalence of metabolic syndrome, while children in low-income countries had the highest.[39] In addition, countries in the stable and decreasing groups performed worse than those in the increasing group in terms of country development, healthcare expenditure, and essential health services coverage. Thus, the fight against liver cancer is extremely challenging in these socially deprived areas, as these countries are not able to ensure sufficient resources for containing the current increasing trends in non-infectious factors as well as eliminating infectious factors that are still prevalent. How can we deal with the double burden of infection-related and non-infection-related liver cancers in a resource-constrained setting? Prioritizing cancer prevention can substantially reduce the number of new cases and deaths through cost-effective interventions. Such actions include the implementation of HBV vaccination programs, continued control of aflatoxin exposure, blood safety, direct-acting antiviral HCV therapy, reduction in alcohol consumption, controlling overweight status, and screening in high-risk individuals.
The strength of this study is that it provides a broad and succinct landscape of global disparities in liver cancer burden across countries over the past three decades using data-driven methods for the first time. Compared to previous studies that have described the global trend of the liver cancer burden,[13,22,40,41] our study further refined our understanding by focusing on the homogeneity and heterogeneity of changes in the liver cancer burden across countries. Moreover, the driving forces underlying this divergence in trends were explored at the risk factor and socioeconomic determinant levels. Finally, projections of future liver cancer burdens at national and regional levels were made. However, some limitations still need to be noted: First, the accuracy of the GBD estimates relies heavily on the quality and quantity of data used in the modeling, which is inconsistent across countries. Although great efforts in data seeking, data processing correction, and modeling have been made in GBD studies, our findings should be interpreted with caution in some data-sparse and often resource-limited locations. Second, as the GMMs are based merely on the data and do not apply a priori hypothesis to the groups, criticisms may point out that the group-based method oversimplifies the true variability of the longitudinal data. However, if the diagnostic criteria related to the analyses are strictly followed, as in our study, GMMs are adequate to model the changing patterns of liver cancer burden and effectively differentiate countries into meaningful groups. Third, because of the lack of exposure data, the contribution of aflatoxin could not be separated from that of HBV, although aflatoxin control was a key driver of the decline in liver cancer incidence in China, Southeast Asia, and sub-Saharan Africa.
In summary, the temporal trends in liver cancer rates have varied globally over the past three decades. Three distinct trajectories were observed: increasing, stable, and decreasing. HBV was the major driver in the decreasing group, whereas alcohol use, HCV, and HBV contributed the most to the trend in the increasing group. Great variations in disease burden are predicted to continue over the next decade, with a disproportionate burden in resource-limited settings. Many more targeted strategies should be established according to the profile of local risk factors.
Data availability statement
Publicly available datasets were used for this study. These can be found in Institute for Health Metrics and Evaluation websites at https://ghdx.healthdata.org/gbd-2019.
Funding
This work was supported by grants from the National Natural Science Foundation of China (Nos. 82273721, 81974492) and the Sanming Project of Medicine in Shenzhen (No. SZSM201911015).
Conflicts of interest
None.
References
1. International Agency for Research on Cancer. Global Cancer Observatory: Cancer Today, 2020. Available from:
https://gco.iarc.fr/today. [Last accessed on November 18, 2021].
2. Kocarnik JM, Compton K, Dean FE, Fu W, Gaw BL, et alGlobal Burden of Disease 2019 CollaborationCancer. incidenceCancer, mortality, years of life lost, years lived with disability, and disability-adjusted life years for 29 cancer groups from 2010 to 2019: A systematic analysis for the Global Burden of Disease Study 2019. JAMA Oncol 2022;8: 420–444. doi: 10.1001/jamaoncol.2021.6987.
3. McGlynn KA, Petrick JL, London WT. Global epidemiology of hepatocellular carcinoma: An emphasis on demographic and regional variability. Clin Liver Dis 2015;19: 223–238. doi: 10.1016/j.cld.2015.01.001.
4. Valery PC, Laversanne M, Clark PJ, Petrick JL, McGlynn KA, Bray F. Projections of primary liver cancer to 2030 in 30 countries worldwide. Hepatology 2018;67: 600–611. doi: 10.1002/hep.29498.
5. Schweitzer A, Horn J, Mikolajczyk RT, Krause G, Ott JJ. Estimations of worldwide prevalence of chronic hepatitis B virus infection: A systematic review of data published between 1965 and 2013. Lancet 2015;386: 1546–1555. doi: 10.1016/s0140-6736(15)61412-x.
6. Blach S, Zeuzem S, Manns M, Altraif I, Duberg AS, et al.Polaris Observatory HCV Collaborators Global prevalence and genotype distribution of hepatitis C virus infection in 2015: A modelling study. Lancet Gastroenterol Hepatol 2017;2: 161–176. doi: 10.1016/s2468-1253(16)30181-9.
7. Griswold MG, Fullman N, Hawley C, Arian N, Zimsen SRM, et al.GBD 2016 Alcohol Collaborators Alcohol use and burden for 195 countries and territories, 1990–2016: A systematic analysis for the Global Burden of Disease Study 2016. Lancet 2018;392: 1015–1035. doi: 10.1016/s0140-6736(18)31310-2.
8. Tanaka H, Imai Y, Hiramatsu N, Ito Y, Imanaka K, Oshita M, et al. Declining incidence of hepatocellular carcinoma in Osaka, Japan, from 1990 to 2003. Ann Intern Med 2008;148: 820–826. doi: 10.7326/0003-4819-148-11-200806030-00004.
9. Wong MCS, Huang JLW, George J, Huang J, Leung C, Eslam M, et al. The changing epidemiology of liver diseases in the Asia-Pacific region. Nat Rev Gastroenterol Hepatol 2019;16: 57–73. doi: 10.1038/s41575-018-0055-0.
10. Bertuccio P, Turati F, Carioli G, Rodriguez T, La Vecchia C, Malvezzi M, et al. Global trends and predictions in hepatocellular carcinoma mortality. J Hepatol 2017;67: 302–309. doi: 10.1016/j.jhep.2017.03.011.
11. Huang DQ, Singal AG, Kono Y, Tan DJH, El-Serag HB, Loomba R. 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. doi: 10.1016/j.cmet.2022.05.003.
12. Lin L, Yan L, Liu Y, Qu C, Ni J, Li H. The burden and trends of primary liver cancer caused by specific etiologies from 1990 to 2017 at the global, regional, national, age, and sex level results from the Global Burden of Disease Study 2017. Liver Cancer 2020;9: 563–582. doi: 10.1159/000508568.
13. Liu Y, Zheng J, Hao J, Wang RR, Liu X, Gu P, et al. Global burden of primary liver cancer by five etiologies and global prediction by 2035 based on Global Burden of Disease Study 2019. Cancer Med 2022;11: 1310–1323. doi: 10.1002/cam4.4551.
14. Davies JI, Reddiar SK, Hirschhorn LR, Ebert C, Marcus ME, Seiglie JA, et al. Association between country preparedness indicators and quality clinical care for cardiovascular disease
risk factors in 44 lower- and middle-income countries: A multicountry analysis of survey data. PLoS Med 2020;17: e1003268. doi: 10.1371/journal.pmed.1003268.
15. Li W, Kemos P, Salciccioli JD, Marshall DC, Shalhoub J, Alazawi W. Socioeconomic factors associated with liver-related mortality from 1985 to 2015 in 36 developed countries. Clin Gastroenterol Hepatol 2021;19: 1698–1707.e13. doi: 10.1016/j.cgh.2020.08.047.
16. Global Burden of Disease Collaborative Network. Global Burden of Disease Study 2019 (GBD 2019) Results. Seattle, United States: Institute for Health Metrics and Evaluation (IHME), 2020. Available from:
http://ghdx.healthdata.org/gbd-results-tool. [Lat accessed on Feburary 4, 2023].
17. World Health Organization. Global Health Observatory data repository. Available from:
http://apps.who.int/gho/data/?theme1/4main. [Last accessed on May 24, 2022].
18. BankWorld. World Bank Open Data. Available from:
https://data.worldbank.org. [Last accessed on May 24, 2022].
19. Nguena Nguefack HL, Pagé MG, Katz J, Choinière M, Vanasse A, Dorais M, et al. Trajectory modelling techniques useful to epidemiological research: A comparative narrative review of approaches. Clin Epidemiol 2020;12: 1205–1222. doi: 10.2147/CLEP.S265287.
20. Riebler A, Held L. Projecting the future burden of cancer: Bayesian age-period-cohort analysis with integrated nested Laplace approximations. Biom J 2017;59: 531–549. doi: 10.1002/bimj.201500263.
21. Knoll M, Furkel J, Debus J, Abdollahi A, Karch A, Stock C. An R package for an integrated evaluation of statistical approaches to cancer incidence projection. BMC Med Res Methodol 2020; 20: 257. doi: 10.1186/s12874-020-01133-5.
22. Liu Z, Xu K, Jiang Y, Cai N, Fan J, Mao X, et al. Global trend of aetiology-based primary liver cancer incidence from 1990 to 2030: A modelling study. Int J Epidemiol 2021;50: 128–142. doi: 10.1093/ije/dyaa196.
23. Smith TR, Wakefield J. A review and comparison of age-period-cohort models for cancer incidence. Statist Sci 2016;31: 591–610. doi: 10.1214/16-sts580.
24. Proust-Lima C, Philipps V, Liquet B. Estimation of extended mixed models using latent classes and latent processes: The R package lcmm. J Stat Softw 2017;78: 1–56. doi: 10.18637/jss.v078.i02.
25. Huang J, Lok V, Ngai CH, Chu C, Patel HK, Thoguluva Chandraseka V, et al. Disease burden,
risk factors, and recent trends of liver cancer: A global country-level analysis. Liver Cancer 2021; 10: 330–345. doi: 10.1159/000515304.
26. Wu J, Yang S, Xu K, Ding C, Zhou Y, Fu X, et al. Patterns and trends of liver cancer incidence rates in Eastern and Southeastern Asian Countries (1983-2007) and predictions to 2030. Gastroenterology 2018;154: 1719–1728.e5. doi: 10.1053/j.gastro.2018.01.033.
27. Liu J, Liang W, Jing W, Liu M. Countdown to 2030: Eliminating hepatitis B disease, China. Bull World Health Organ 2019; 97: 230–238. doi: 10.2471/BLT.18.219469.
28. Ryerson AB, Eheman CR, Altekruse SF, Ward JW, Jemal A, Sherman RL, et al. Annual Report to the Nation on the Status of Cancer, 1975-2012, featuring the increasing incidence of liver cancer. Cancer 2016;122: 1312–1337. doi: 10.1002/cncr.29936.
29. Akinyemiju T, Abera S, Ahmed M, Alam N, Alemayohu MA, et al.Global Burden of Disease Liver Cancer Collaboration The burden of primary liver cancer and underlying etiologies from 1990 to 2015 at the global, regional, and national level: Results from the Global Burden of Disease Study 2015. JAMA Oncol 2017; 3: 1683–1691. doi: 10.1001/jamaoncol.2017.3055.
30. Inoue Y, Qin B, Poti J, Sokol R, Gordon-Larsen P. Epidemiology of obesity in adults: Latest trends. Curr Obes Rep 2018;7: 276–288. doi: 10.1007/s13679-018-0317-8.
31. Morgan TR, Mandayam S, Jamal MM. Alcohol and hepatocellular carcinoma. Gastroenterology 2004;127(5 Suppl 1): S87–S96. doi: 10.1053/j.gastro.2004.09.020.
32. Saitta C, Pollicino T, Raimondo G. Obesity and liver cancer. Ann Hepatol 2019;18: 810–815. doi: 10.1016/j.aohep.2019.07.004.
33. Ott JJ, Horn J, Krause G, Mikolajczyk RT. Time trends of chronic HBV infection over prior decades–A global analysis. J Hepatol 2017;66: 48–54. doi: 10.1016/j.jhep.2016.08.013.
34. Hassanin A, Kamel S, Waked I, Fort M. Egypt's ambitious strategy to eliminate hepatitis C virus: A case study. Glob Health Sci Pract 2021;9: 187–200. doi: 10.9745/GHSP-D-20-00234.
35. Fan JG, Kim SU, Wong VW. New trends on obesity and NAFLD in Asia. J Hepatol 2017;67: 862–873. doi: 10.1016/j.jhep.2017.06.003.
36. World Health Organization. Global health sector strategy on viral hepatitis 2016-2021, 2016. Available from:
https://www.who.int/publications/i/item/WHO-HIV-2016.06. [Last accessed on July 27, 2022].
37. Petrick JL, Florio AA, Znaor A, Ruggieri D, Laversanne M, Alvarez CS, et al. International trends in hepatocellular carcinoma incidence, 1978-2012. Int J Cancer 2020; 147: 317–330. doi: 10.1002/ijc.32723.
38. Manthey J, Shield KD, Rylett M, Hasan OSM, Probst C, Rehm J. Global alcohol exposure between 1990 and 2017 and forecasts until 2030: A modelling study. Lancet 2019;393: 2493–2502. doi: 10.1016/s0140-6736(18)32744-2.
39. Noubiap JJ, Nansseu JR, Lontchi-Yimagou E, Nkeck JR, Nyaga UF, Ngouo AT, et al. Global, regional, and country estimates of metabolic syndrome burden in children and adolescents in 2020: A systematic review and modelling analysis. Lancet Child Adolesc Health 2022;6: 158–170. doi: 10.1016/s2352-4642(21)00374-6.
40. Bray F, Jemal A, Grey N, Ferlay J, Forman D. Global cancer transitions according to the human development index (2008-2030): A population-based study. Lancet Oncol 2012;13: 790–801. doi: 10.1016/s1470-2045(12)70211-5.
41. Choi S, Kim BK, Yon DK, Lee SW, Lee HG, Chang HH, et al. Global burden of primary liver cancer and its association with underlying aetiologies, sociodemographic status, and sex differences, 1990-2019: A DALYs-based analysis of the Global Burden of Disease 2019 study. Clin Mol Hepatol 2023. doi: 10.3350/cmh.2022.0316.
