ABSTRACT
-
Background/Aims
The association between aspirin use and hepatocellular carcinoma (HCC) risk in patients with metabolic dysfunction-associated steatotic liver disease (MASLD) remains unclear. This study evaluated the effect of aspirin on HCC development in MASLD patients using Korean National Health Insurance Service (NHIS) and UK Biobank (UKB) databases.
-
Methods
A retrospective cohort analysis was conducted using the NHIS database with a 3-year landmark design. Baseline characteristics were balanced using inverse probability of treatment weighting (IPTW) and 1:3 propensity score matching (PSM). Additionally, Mendelian randomization analysis was performed in the UKB cohort using a genomic risk score (GRS) for salicylic acid, based on genetic variants related to aspirin metabolism, as a proxy for aspirin use.
-
Results
In the NHIS cohort, 6,584,155 eligible patients were included, of whom 1,723,435 had MASLD. After PSM, aspirin use was associated with a significantly lower risk of HCC compared to no aspirin use, in both the overall population (adjusted subdistribution hazard ratio [ASHR]=0.86; 95% confidence interval [CI] 0.78–0.95; P=0.002) and MASLD group (ASHR=0.86; 95% CI 0.75–0.99; P=0.036). Similar results were reproduced in the IPTW population and several sensitivity and subgroup analyses. In the UKB cohort, individuals in the top 95% of GRS had a significantly lower risk of HCC compared to those in the bottom 5%, in both the overall population (ASHR=0.61; 95% CI 0.39–0.95; P=0.028) and MASLD group (ASHR=0.47; 95% CI 0.29–0.76; P=0.002).
-
Conclusions
Findings from both population-based and genetic analyses suggest a possible protective association between aspirin use and HCC in patients with MASLD, which warrants further validation.
-
Keywords: Fatty liver; Aspirin; Salicylic acid; Liver cancer; UK Biobank
Study Highlights
• This study used two large-scale cohorts to evaluate the association between aspirin and HCC risk in MASLD.
• Using a 3-year landmark retrospective design in a Korean nationwide cohort, aspirin use was consistently associated with a lower risk of HCC in both the overall population and the MASLD subgroup.
• Complementary genetic analyses employing a salicylic acid-related genomic risk score as a proxy for aspirin exposure showed concordant protective associations in the UK Biobank.
• Collectively, results from population-based and genetic analyses support a potential chemopreventive role of aspirin against HCC in patients with MASLD.
Graphical Abstract
INTRODUCTION
Metabolic dysfunction-associated steatotic liver disease (MASLD) was recently proposed to emphasize the importance of metabolic disorders in patients with steatotic liver disease, which is the most common cause of chronic liver disease worldwide [
1-
3]. Although only a small proportion of MASLD patients progress to liver cirrhosis (LC) and hepatocellular carcinoma (HCC), the large absolute number of patients with MASLD results in a substantial disease burden [
4]. As MASLD-related HCC continues to increase globally, there is an urgent need to establish appropriate management and preventative measures [
5].
While it is well established that antiviral therapy can effectively prevent the development of HCC in patients with chronic hepatitis B (CHB) [
6], there is a lack of data on medication that can reduce the incidence of HCC in MASLD patients. Recently, several drugs have been developed for patients with MASLD or metabolic dysfunction-associated steatohepatitis (MASH) [
7,
8]. However, these novel agents have been validated in randomized controlled trials (RCTs) whose primary outcomes are improvement of steatohepatitis and fibrosis on biopsy tissue, not the development of HCC. It is not feasible to confirm in an RCT that a specific agent prevents HCC, as only a small proportion of patients with MASLD and MASH progress to HCC, and it generally takes a significant amount of time for these populations to develop HCC.
Meanwhile, several studies have shown that aspirin use is associated with a lower risk of HCC in patients with chronic liver disease [
9-
19]. However, most of these studies have been conducted among patients with chronic viral hepatitis, and there is a lack of evidence confirming its effectiveness in the MASLD population. A previous study demonstrated that platelets play a pivotal role in MASH and hepatocarcinogenesis, and that antiplatelet therapy, including aspirin, can effectively inhibit HCC development [
20]. In addition, a recent RCT showed that 6 months of low-dose aspirin treatment reduced hepatic fat content in patients with MASLD [
21].
It is difficult to determine whether aspirin has a protective effect against HCC development in the general MASLD population through RCTs. Therefore, in this study, we aimed to investigate the effect of aspirin use in MASLD patients utilizing two large, complementary population-based cohorts. First, we performed a population-based cohort analysis using nationwide claims data from Korea to confirm whether aspirin use was associated with a lower incidence of HCC in patients with MASLD. In parallel, a Mendelian randomization (MR) analysis was conducted in the UK Biobank (UKB) cohort to investigate whether genetic variants corresponding to aspirin use were related to HCC incidence.
MATERIALS AND METHODS
Population-based cohort study
Study population and study design
This study included individuals aged 30 or older who received the national health examination provided by the National Health Insurance Service (NHIS) of Korea in 2009, with the examination date designated as the cohort entry date. To avoid immortal time bias, we implemented a 3-year landmark analysis, setting the index date as three years after the cohort entry date. This 3-year landmark period was selected based on the evidence from previous studies [
13,
22,
23]. Participants were excluded based on the following criteria: (i) consecutive aspirin use for more than 90 days before the cohort entry date; (ii) aspirin use within 6 months after the cohort entry date; (iii) clopidogrel use for over 90 days before the index date; (iv) liver-related diseases (details provided in
Supplementary Table 1) prior to the index date; (v) incident liver-related diseases within 6 months after the index date; (vi) a history of acquired immune deficiency syndrome before the index date; and (vii) incomplete data. Further details on study design are presented in
Figure 1.
Patients with MASLD were identified based on the presence of hepatic steatosis and more than one metabolic risk factor without excessive alcohol consumption. Details on metabolic risk factors are provided in the
Supplementary Methods. Patients with MetALD (MASLD with increased alcohol intake) were explicitly excluded from the MASLD cohort. Excessive alcohol consumption was defined as any type of alcohol with ≥30 g/day in men or ≥20 g/day in women.
Aspirin user definition and outcome
To define sustained aspirin use, we set a 90-day threshold to reflect its role in long-term preventive therapy. Aspirin users were defined as patients who were newly prescribed aspirin with at least 90 consecutive days prior to the landmark time [
12]. Those with no aspirin prescriptions, or with prescriptions lasting fewer than 90 consecutive days before the landmark time, were classified as aspirin non-users (
Fig. 1B). For instance, a patient who initiated aspirin 10 days after the landmark time and continued for more than 90 days was still categorized as an aspirin non-user. Patients classified into each treatment group at the landmark time were assumed to remain in that group over the entire follow-up.
Primary outcome was the development of HCC, defined by the the 10th revision of the International Classification of Diseases (ICD-10) code C22 (
Supplementary Table 1). HCC-related death was not included due to insufficient clinical specificity in the cause-of-death data. Negative control outcome was defined as death related to external causes of morbidity and mortality (ICD-10 code V01–Y98). Individuals were censored at the time of HCC diagnosis, all-cause death, or at the end of the follow-up period (December 31, 2019), whichever occurred first.
Statistical analysis
To compare the baseline characteristics, we used t-test on 2 egorical variables in the crude population. Inverse probability of treatment weight (IPTW) was estimated using multivariable logistic regression.
In the crude population, we estimated the incidence rates and their 95% confidence intervals (CIs) using a Poisson distribution. To account for competing risks, such as all-cause death or other-cause death, we applied the Fine and Gray model, and crude subdistribution hazard ratios (SHRs) were estimated. Two sets of covariates were considered to demonstrate how effect estimates change with the inclusion of additional confounders; Model 1 adjusted for age and sex, and Model 2 adjusted for all potential confounders.
To confirm the validity of our findings, we conducted multiple sensitivity analyses, including a landmark-based sequential trial design (
Supplementary Table 2), and additional methodological details are described in the
Supplementary Methods.
Genetic instruments for aspirin metabolism
Single nucleotide polymorphisms (SNPs) associated with aspirin metabolism were identified from previous genome-wide association studies (GWASs) on salicylic acid levels [
24,
25]. A genetic risk score (GRS) for the salicylic acid levels was calculated using SNPs that satisfied the criteria for valid instrument variables. The causal effect of salicylic acid on the risk of HCC was assessed by these instruments.
RESULTS
Population-based cohort study
Baseline characteristics
We identified a total of 6,584,155 eligible patients after applying the exclusion criteria (
Fig. 1A). Among them, 1,723,435 individuals had MASLD.
Table 1 shows baseline characteristics of the crude population. Most of the covariates were significantly different in both the overall population and MASLD group. The weighted baseline characteristics of the IPTW population are presented in
Supplementary Table 3. Using stabilized weights, most confounders remained well balanced. Similarly, both groups were balanced in the propensity score matching (PSM) population (
Supplementary Table 4).
Risk of hepatocellular carcinoma
Table 2 presents the risk of HCC in the crude and IPTW or PSM populations. In the crude population, univariable analysis showed that aspirin use was associated with a higher risk of HCC compared to no aspirin use in the overall population (aspirin use vs. no aspirin use: SHR=1.92; 95% CI 1.76–2.09;
P<0.001;
Fig. 2A) and in the MASLD group (SHR=1.67; 95% CI 1.47–1.89;
P<0.001;
Fig. 2B). However, after applying IPTW, the aspirin user group showed a significantly lower risk of HCC in the overall population (adjusted SHR [ASHR]=0.64; 95% CI 0.48–0.84;
P=0.001;
Fig. 2C). In the IPTW population with MASLD, aspirin use demonstrated a trend toward a lower risk of HCC, although not statistically significant (ASHR=0.77; 95% CI 0.57–1.04;
P=0.090;
Fig. 2D). After applying the 1:3 PSM, aspirin use was associated with a significantly lower risk of HCC compared to no aspirin use in the overall population (ASHR=0.86; 95% CI 0.78–0.95;
P=0.002;
Fig. 2E) and in the MASLD group (ASHR=0.86; 95% CI 0.75–0.99;
P=0.036;
Fig. 2F).
Sensitivity and subgroup analyses
Similar results were reproduced when using different landmark years (
Supplementary Table 5). A lower risk of HCC in aspirin users was observed when hepatic steatosis was defined as HSI ≥36 or when the 6-month washout period was not considered (
Supplementary Table 6). There was no difference in HCC risk between aspirin users and non-users after IPTW in both MASLD patients with fatty liver index (FLI) ≥60 (ASHR=1.01; 95% CI 0.54–0.91;
P=0.965) and MetALD patients (ASHR=1.29; 95% CI 0.59–2.80;
P=0.527).
Supplementary Table 7 presents the results of subgroup analyses. Before IPTW adjustment, the interaction effects between aspirin use and several confounders on HCC outcome were not statistically significant in the MASLD group, except for sex and diabetes mellitus (DM). In the negative control outcome analyses, aspirin use did not show a significant effect, suggesting that the primary results were not confounded by the residual biases (
Supplementary Table 8). Lastly, all analyses from the landmark-based sequential trial yielded statistically significant pooled odds ratios in the crude, IPTW, and PSM populations (
Supplementary Table 9), as well as in subgroups restricted to patients without concomitant liver disease or without both concomitant liver disease and cirrhosis (
Supplementary Table 10).
GRS-based one-sample MR
Six SNPs (rs8062555, rs9922093, rs146980165, rs7499557, rs7500194239, and rs8056693) related to the aspirin metabolism pathway were selected as genetic instruments (
Supplementary Table 11). These SNPs were all located within or near the ACSM2B gene, which was known to be involved in the metabolism of aspirin metabolites [
26]. Among them, two SNPs were found to have strict genome-wide significance. These genetic instruments strongly predicted higher salicylic acid levels (F-statistics=29.99). Then, we divided the patients in the UKB population based on GRS percentiles and compared the risk of HCC. The two groups were well-balanced at baseline (
Supplementary Table 12). The top 95% of patients exhibited a significantly lower risk of HCC (Gray’s test
P=0.03;
Supplementary Fig. 1A). The top 95% of patients had a significantly lower risk of HCC than the bottom 5% of patients after adjusting for age, sex, and the first 10 principal components of genetic factors (ASHR=0.62; 95% CI 0.40–0.96;
P=0.034). This relationship remained significant after adjusting for age, sex, the first 10 principal components, LC, DM, hypertension, body mass index (BMI), and waist circumference (ASHR=0.61; 95% CI 0.39–0.95;
P=0.028). Similar results were confirmed when patients were stratified by the 10 percentiles of GRS instead of the 5 percentiles (Gray’s test
P=0.03;
Supplementary Fig. 1B).
HCC risk in patients with MASLD
We selected patients with MASLD (n=239,719) in the UKB cohort, and divided the patients based on the GRS (
Supplementary Table 13). The risk of HCC was compared among patients based on their GRS percentiles.
Supplementary Figure 2A shows that the top 95% of patients had a significantly lower risk of HCC compared to the bottom 5% of patients (Gray’s test
P=0.006). The top 95% of patients exhibited a significantly lower risk of HCC than the bottom 5% of patients after adjusting for age, sex, and the first 10 principal components of genetic factors (ASHR=0.52; 95% CI 0.33–0.84;
P=0.007). Similar results were maintained after adjusting for age, sex, the first 10 principal components, LC, DM, hypertension, BMI, and waist circumference (ASHR=0.47; 95% CI 0.29–0.76;
P=0.002). Stratifying patients by the 10 percentiles of GRS instead of the 5 percentiles did not change the main findings (Gray’s test
P=0.049;
Supplementary Fig. 2B).
DISCUSSION
In this study using two complementary population-based cohorts, aspirin use was associated with a lower risk of HCC in both the general population and patients with MASLD. In a population-based cohort analysis in the Korean NHIS database, the aspirin user group was associated with a significantly lower incidence of HCC compared to the aspirin non-user group after applying IPTW and PSM. A one-sample MR analysis was performed in the UKB database by calculating a GRS for salicylic acid levels as a genetic proxy for the effect of aspirin use, and a significantly lower incidence of HCC was confirmed in patients with a high GRS. Our findings suggest that aspirin may have a protective effect against the development of HCC in patients with MASLD.
Previous experimental studies have shown the chemopreventive effect of aspirin against the development of HCC. Platelets are known to recruit circulating CD8
+ T cells and induce the release of platelet-derived growth factors or cytokines [
27,
28]. Intrahepatic platelet aggregation and activation was observed in animal models of MASLD, leading to additional liver damage, and aspirin treatment inhibited hepatocarcinogenesis by reducing steatosis, atherosclerosis, and hepatic immune cell infiltration [
20]. Aspirin may also suppress HCC development through platelet-independent pathways. Previous studies have reported that aspirin may reduce hepatic fibrogenesis by inhibiting the activation of hepatic stellate cells and attenuating the accumulation of abnormal extracellular matrix [
29,
30]. In addition to animal studies, a recent RCT in patients with MASLD confirmed that aspirin significantly reduced hepatic fat compared to placebo [
21]. Patients treated with low-dose (81 mg) aspirin daily for 6 months had a significant reduction in hepatic fat content measured by proton magnetic resonance spectroscopy or magnetic resonance imaging proton density fat fraction compared to those who received placebo [
21].
Since it is practically unfeasible to conduct an RCT to confirm the protective effect of aspirin against HCC development in the MASLD population, we aimed to verify our study hypothesis by using two independent population-based cohorts and applying different analytical methods. Our study is distinct from previous NAFLD-based research [
18], as it evaluates the effect of aspirin using the new MASLD criteria, which emphasize metabolic dysfunction as a key pathogenic driver. By adopting the recent reclassification of steatotic liver disease, our study offers novel clinical evidence within the updated framework. In the cohort analysis using the Korean nationwide database, the protective effect of aspirin against HCC was statistically significant in all groups except in the IPTW population with MASLD (0.05<
P<0.10). In the unadjusted analysis, aspirin use appeared to be associated with an increased risk of HCC, which may be confounded by baseline metabolic risk differences between users and non-users. After adjustment with IPTW and PSM, however, aspirin use was associated with a significantly lower risk of HCC.
The protective effect of aspirin was not confirmed in patients with FLI ≥60 or MetALD. A recent nationwide cohort study conducted in Taiwan showed that patients who took aspirin for more than 90 days had a significantly lower risk of HCC compared with the control group [
18], which is consistent with our findings. In that study, the longer the duration of aspirin use, the greater the protective effect against HCC, and in high-risk groups (aged over 55 years with elevated liver enzymes), the magnitude of the risk reduction decreased. In another study investigating the effect of aspirin in patients with CHB, patients without baseline LC had a significantly lower risk of HCC in the aspirin user group compared with the control group [
19]. However, this association was not confirmed in patients with baseline LC. These findings collectively suggest that aspirin may be less effective in preventing HCC in patients with established or progressive hepatic fibrosis. As the protective effect of aspirin against HCC is primarily attributed to its ability to reduce hepatic inflammation, it is reasonable to assume that this effect would be more pronounced in patients with early-stage liver disease compared to those with advanced cirrhosis. Moreover, individuals with FLI ≥60 may have more frequent metabolic dysfunction, which could increase the risk of CVD as a competing risk and thereby attenuate the relative incidence of hepatic events observed in our study.
To utilize the effects of aspirin as a genetic instrument in MR analysis, we identified SNPs associated with aspirin metabolism from two recent GWAS studies. Aspirin is metabolized into its primary active metabolite, salicylic acid, which exerts various biochemical effects. Therefore, to comprehensively capture the diverse actions of aspirin, we used summary statistics from a GWAS on salicylic acid. Consequently, we assumed that SNPs associated with serum salicylic acid levels could serve as proxies for mimicking the pharmacological effects of aspirin. In this context, MR studies on drug effects have recently garnered significant attention [
31], and are regarded as valuable methodologies to minimize the impact of confounders, especially in cases where conducting an RCT is not feasible [
32]. Unlike conventional observational analyses, this genetic approach enhances causal inference and offers novel insights into the potential protective effect of aspirin against HCC [
33].
The three core assumptions of MR were evaluated and considered to be appropriately satisfied. The selected SNPs demonstrating strong associations with salicylic acid metabolism (relevance assumption) were located near the ACSM2B gene, which is known to mediate aspirin metabolism, reducing the likelihood of horizontal pleiotropy (exclusion restriction). In addition, the potential for confounding by environmental exposures such as dietary benzoates was considered minimal (independence). Genetic polymorphisms in aspirin metabolism pathways were assumed to reflect variation in salicylic acid clearance. Accordingly, a lower GRS based on these SNPs was interpreted as indicative of higher metabolic activity and reduced circulating salicylic acid levels. Based on prior clinical studies demonstrating that serum salicylic acid levels increase with aspirin intake and are minimally affected by dietary sources, this genetically predicted variation in salicylic acid levels was used as a proxy for aspirin exposure and effect [
34,
35].
In this study, the protective effect of aspirin against HCC was confirmed in two independent cohorts of different ethnicities. Furthermore, we tried to compensate for the inherent limitations of the population-based cohort study by additional MR analysis and negative control outcome analyses. However, there are several limitations in the current study. First, despite multiple adjustments and sensitivity analyses, this study may be subject to selection bias and residual confounding. Approximately 15% of individuals were excluded due to missing data, which may introduce selection bias. Furthermore, the use of a landmark design and the exclusion of post-index liver disease events may also lead to selection bias. This follow-up window was implemented to account for potential diagnostic delay, but residual confounding cannot be entirely ruled out [
36]. Second, actual compliance was not considered in aspirin users, and over-the-counter use could not be assessed in non-users. These limitations may have attenuated the observed differences in HCC risk. Moreover, the dose–response relationship should also be evaluated in future studies. Third, hepatic steatosis was assessed using non-invasive markers rather than liver biopsy or imaging in this study. Although FLI has been validated as a reasonable surrogate in previous studies [
37-
39], the use of an FLI ≥30 threshold in a general health screening population may have contributed to the low prevalence of diabetes in the MASLD population. This threshold likely identified relatively healthy individuals, which may explain the attenuated protective effect of aspirin when using the FLI ≥60 cutoff. Fourth, the use of a genetic proxy between aspirin use and salicylic acid levels warrants further validation, since ACSM2B is also involved in other metabolic pathways that could confound the results [
26]. In addition, because aspirin use in the UKB cohort was based on self-reported data without reliable information on cumulative duration or prescription records, we were unable to perform a robust subgroup analysis restricted to aspirin users. Analyses confined to aspirin users might have strengthened the findings of the GRS analysis and provided greater reliability. This limitation should therefore be considered when interpreting the genetic risk analysis, as it may have led to attenuation of the observed associations. Fifth, the use of ICD-10 code C22 may have included non-HCC liver cancers; however, aspirin use has not been strongly associated with these malignancies [
40]. Therefore, any potential dilution of the association between aspirin use and HCC is likely minimal, while ensuring that HCC cases were adequately captured. Sixth, another important limitation concerns the prespecified choice of the 3-year landmark time. Sensitivity analyses using 2- and 5-year landmarks did not fully align with the 3-year results. However, the 3-year landmark was selected to reflect the long latency of HCC development in the MASLD population and the cumulative chemopreventive effect of aspirin [
13,
22,
23]. More sophisticated analytic frameworks to mitigate immortal time bias such as time-varying exposure models or target trial emulation with a clone-censor-weight approach may provide a more robust evaluation. Lastly, the potential risk of bleeding associated with aspirin use was not evaluated. A recent meta-analysis found that in patients with chronic liver disease, aspirin use reduced the risk of HCC by 30% but increased the risk of major bleeding by 10% [
15]. In another study of patients with CHB [
19], aspirin use was associated with a higher risk of bleeding in those without baseline LC, whereas no excess risk was observed among patients with LC. Although our data shows that aspirin use is associated with a lower risk of HCC in the general population, it is unlikely that aspirin would be recommended for all patients given the bleeding risk. Further studies are needed to balance risks and benefits, and to identify the patients most likely to benefit.
In conclusion, we demonstrated that aspirin use is associated with a reduced risk of HCC in patients with MASLD by analyzing two independent population-based cohorts and employing diverse analytical methods. Both population-based cohort and MR analyses consistently confirmed the protective effect of aspirin against HCC, and the main results were maintained in multiple sensitivity and subgroup analyses. However, given the several limitations of the current study, these findings should be interpreted as hypothesis-generating, and further large-scale, prospective studies are warranted to validate these findings and to identify the optimal population that could benefit from aspirin therapy in patients with MASLD.
FOOTNOTES
-
Authors’ contributions
Conceptualization: Juhee Ahn, Moon Haeng Hur, Hyunjae Shin, Min Kyung Park, Sungho Won, and Yoon Jun Kim; Data Curation: Juhee Ahn, Moon Haeng Hur, Hyunjae Shin, and Min Kyung Park; Formal analysis: Juhee Ahn, Moon Haeng Hur, Hyunjae Shin, Min Kyung Park, Sungho Won, and Yoon Jun Kim; Methodology: Juhee Ahn, Moon Haeng Hur, Hyunjae Shin, Min Kyung Park, Sungho Won, Jeayeon Park, Yunmi Ko, Youngsu Park, Yun Bin Lee, and Eun Ju Cho; Supervision: Sungho Won, Jeong-Hoon Lee, Su Jong Yu, Jung-Hwan Yoon, and Yoon Jun Kim; Writing - Original Draft: Juhee Ahn, Moon Haeng Hur, and Hyunjae Shin; Writing - Review & Editing: Min Kyung Park, Sungho Won, Jeayeon Park, Yunmi Ko, Youngsu Park, Yun Bin Lee, Eun Ju Cho, Jeong-Hoon Lee, Su Jong Yu, Jung-Hwan Yoon, and Yoon Jun Kim.
-
Acknowledgements
This study was supported by the Research Supporting Program of the Korean Association for the Study of the Liver and the Korean Liver Foundation and research grants from Dong-A ST Co. (Seoul, Korea) and Yuhan Corporation (Seoul, Korea).
The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
The study was based on the data provided by the UKB Consortium (application no. 53799) and the FinnGen Consortium.
-
Conflicts of Interest
Juhee Ahn: Nothing to declare; Moon Haeng Hur: Nothing to declare; Hyunjae Shin: Nothing to declare; Min Kyung Park: Nothing to declare; Sungho Won: Nothing to declare; Jeayeon Park: Nothing to declare; Yunmi Ko: Nothing to declare; Youngsu Park: Nothing to declare; Yun Bin Lee: Receives research grants from Samjin Pharmaceuticals and Yuhan Pharmaceuticals; Eun Ju Cho: Nothing to declare; Jeong-Hoon Lee: Receives research grants from Yuhan Pharmaceuticals and GreenCross Cell, and lecture fees from GreenCross Cell, Daewoong Pharmaceuticals, and Gilead Korea; Su Jong Yu: Receives research grants from Yuhan Pharmaceuticals and Daewoong Pharmaceuticals; Jung-Hwan Yoon: Receives research grants from Bayer, Daewoong Pharmaceuticals, and Bukwang Pharmaceutical; Yoon Jun Kim: Receives research grants from BTG, Boston Scientific, AstraZeneca, Gilead Sciences, Samjin, BL&H, and Bayer, and lecture fees from Roche, Abbvie, Eisai, Boston Scientific, BMS, BTG, Bayer, MSD, Novo Nordisk, Green-Cross Cell, Boehringer Ingelheim, and Gilead Sciences.
SUPPLEMENTARY MATERIAL
Supplementary material is available at Clinical and Molecular Hepatology website (
http://www.e-cmh.org).
Supplementary Figure 1.
Incidence of HCC according to the GRS in the overall UK Biobank population. (A) GRS stratified at the 95 percentiles and (B) GRS stratified at the 90 percentiles. GRS, genomic risk score; HCC, hepatocellular carcinoma.
Supplementary Figure 2.
Incidence of HCC according to the GRS among patients with MASLD in the UK Biobank. (A) GRS stratified at the 95 percentiles and (B) GRS stratified at the 90 percentiles. GRS, genomic risk score; HCC, hepatocellular carcinoma; MASLD, metabolic dysfunction-associated steatotic liver disease.
Figure 1.(A) Patient flow diagram and (B) study design in the Korean nationwide cohort. AIDS, acquired immune deficiency syndrome; IPTW, inverse probability of treatment weighting.
Figure 2.Cumulative incidence of hepatocellular carcinoma in the Korean nationwide cohort according to aspirin use, presented for the crude population ([A] overall, [B] MASLD), the IPTW population ([C] overall, [D] MASLD), and the 1:3 PSM population ([E] overall, [F] MASLD). Differences of cumulative incidence were compared using Gray’s test in overall and PSM populations. IPTW, inverse probability of treatment weighting; MASLD, metabolic dysfunction-associated steatotic liver disease; PSM, propensity score matching.
Table 1.Baseline characteristics of the study cohort
Table 1.
|
Total |
Overall aspirin non-user |
Overall aspirin user |
P-value |
MASLD total |
MASLD aspirin non-user |
MASLD aspirin user |
P-value |
|
(n=6,584,155) |
(n=6,460,945) |
(n=123,210) |
(n=1,723,435) |
(n=1,693,257) |
(n=50,178) |
|
Clinical characteristics |
|
|
|
|
|
|
|
|
|
Age (yr) |
48.3±11.8 |
48.1±11.7 |
60.0±10.3 |
<0.001*
|
49.3±11.9 |
48.9±11.9 |
59.6±10.1 |
<0.001*
|
|
Male |
3,442,553 (52.3) |
3,382,177 (52.3) |
60,376 (49.0) |
<0.001 |
1,245,753 (71.5) |
1,217,460 (71.9) |
28,293 (56.4) |
<0.001 |
|
Household income quartile |
|
|
|
<0.001 |
|
|
|
<0.001 |
|
1Q |
1,253,985 (19.0) |
1,228,596 (19.0) |
25,389 (20.6) |
|
283,724 (16.3) |
273,753 (16.2) |
9,971 (19.9) |
|
|
2Q |
1,375,926 (20.9) |
1,353,139 (20.9) |
22,787 (18.5) |
|
332,438 (19.1) |
323,194 (19.1) |
9,244 (18.4) |
|
|
3Q |
1,826,317 (27.7) |
1,795,701 (27.8) |
30,616 (24.8) |
|
512,320 (29.4) |
499,466 (29.5) |
12,854 (25.6) |
|
|
4Q |
2,127,927 (32.3) |
2,083,509 (32.2) |
44,418 (36.1) |
|
614,953 (35.3) |
596,844 (35.2) |
18,109 (36.1) |
|
|
Residential area |
|
|
|
<0.001 |
|
|
|
<0.001 |
|
Rural |
2,094,130 (31.8) |
2,049,700 (31.7) |
44,430 (36.1) |
|
561,986 (32.2) |
544,116 (32.1) |
17,870 (35.6) |
|
|
Urban |
4,490,025 (68.2) |
4,411,245 (68.3) |
78,780 (63.9) |
|
1,181,449 (67.8) |
1,149,141 (67.9) |
32,308 (64.4) |
|
|
BMI (kg/m2) |
23.6±3.0 |
23.5±3.0 |
24.7±3.0 |
<0.001*
|
26.3±2.5 |
26.3±2.5 |
26.6±2.6 |
<0.001 |
|
Waist circumference (cm) |
79.7±8.5 |
79.6±8.5 |
83.6±8.2 |
<0.001*
|
87.8±6.1 |
87.7±6.1 |
89.0±6.4 |
<0.001*
|
|
Comorbidity |
|
|
|
|
|
|
|
|
|
Diabetes mellitus |
140,691 (2.1) |
127,663 (2.0) |
13,028 (10.6) |
<0.001 |
63,149 (3.6) |
56,880 (3.4) |
6,269 (12.5) |
<0.001 |
|
Hypertension |
745,553 (11.3) |
688,234 (10.7) |
57,319 (46.5) |
<0.001 |
304,773 (17.5) |
278,931 (16.5) |
25,842 (51.5) |
<0.001 |
|
Dyslipidemia |
308,116 (4.7) |
289,848 (4.5) |
18,268 (14.8) |
<0.001 |
132,986 (7.6) |
124,146 (7.3) |
8,840 (17.6) |
<0.001 |
|
Chronic kidney disease |
767,214 (11.7) |
735,878 (11.4) |
31,336 (25.4) |
<0.001 |
120,007 (6.9) |
111,537 (6.6) |
8,470 (16.9) |
<0.001 |
|
Viral hepatitis |
149,851 (2.3) |
146,850 (2.3) |
3,001 (2.4) |
<0.001 |
41,413 (2.4) |
40,171 (2.4) |
1242 (2.5) |
0.136 |
|
Liver cirrhosis |
12,131 (0.2) |
11,848 (0.2) |
283 (0.2) |
<0.001 |
3,697 (0.2) |
3,575 (0.2) |
122 (0.2) |
0.125 |
|
Prior heart failure |
1,169 (0.0) |
1,068 (0.0) |
101 (0.1) |
<0.001 |
431 (0.0) |
396 (0.0) |
35 (0.1) |
0.026 |
|
Prior myocardial infarction |
517 (0.0) |
490 (0.0) |
27 (0.0) |
<0.001 |
176 (0.0) |
166 (0.0) |
10 (0.0) |
<0.001 |
|
Prior stroke |
2,709 (0.0) |
2,474 (0.0) |
235 (0.2) |
<0.001 |
935 (0.1) |
825 (0.0) |
110 (0.2) |
<0.001 |
|
Laboratory findings |
|
|
|
|
|
|
|
|
|
Systolic BP (mmHg) |
121.5±14.4 |
121.3±14.3 |
130.4±15.7 |
<0.001*
|
126.3±13.8 |
126.1±13.7 |
132.2±15.2 |
<0.001*
|
|
Diastolic BP (mmHg) |
75.9±9.7 |
75.8±9.7 |
80.2±10.2 |
<0.001*
|
79.0±9.4 |
79.0±9.3 |
81.4±10.0 |
<0.001*
|
|
Hemoglobin (g/dL) |
13.9±1.6 |
13.9±1.6 |
13.8±1.5 |
<0.001*
|
14.5±1.5 |
14.5±1.5 |
14.1±1.5 |
<0.001 |
|
AST (IU/L) |
23.3±6.8 |
23.3±6.8 |
24.7±7.2 |
<0.001*
|
25.6±7.3 |
25.5±7.2 |
25.9±7.6 |
<0.001*
|
|
ALT (IU/L) |
22.2±11.0 |
22.2±11.0 |
23.5±10.8 |
<0.001*
|
29.2±12.7 |
29.2±12.7 |
27.3±11.9 |
<0.001*
|
|
Total cholesterol (mg/dL) |
196.4±35.1 |
196.3±35.0 |
203.1±37.7 |
<0.001*
|
207.2±35.7 |
207.1±35.6 |
208.4±38.5 |
<0.001*
|
|
Triglycerides (mg/dL) |
124.7±70.1 |
124.4±69.9 |
141.9±73.2 |
<0.001*
|
182.8±77.0 |
182.9±77.0 |
180.1±76.6 |
<0.001 |
|
HDLc (mg/dL) |
55.0±13.5 |
55.1±13.5 |
53.1±13.2 |
<0.001*
|
49.4±11.8 |
49.4±11.7 |
49.9±12.3 |
<0.001*
|
|
LDLc (mg/dL) |
116.2±32.3 |
116.1±32.3 |
121.3±35.2 |
<0.001*
|
121.2±34.2 |
121.1±34.1 |
122.3±36.4 |
<0.001*
|
|
Fasting glucose (mg/dL) |
93.9±13.2 |
93.8±13.1 |
99.4±16.1 |
<0.001*
|
97.2±14.4 |
97.0±14.4 |
101.8±16.6 |
<0.001*
|
|
Creatinine (mg/dL) |
0.9±0.2 |
0.9±0.2 |
0.9±0.2 |
<0.001*
|
1.0±0.2 |
1.0±0.2 |
0.9±0.2 |
<0.001*
|
|
GFR (mL/min/1.73 m2) |
85.1±23.5 |
85.3±23.5 |
74.7±22.8 |
<0.001*
|
93.6±25.1 |
93.9±25.1 |
80.7±23.4 |
<0.001*
|
|
BARD score |
1.8±0.8 |
1.8±0.7 |
1.9±0.8 |
<0.001*
|
1.6±1.0 |
1.6±1.0 |
1.9±1.0 |
<0.001*
|
|
Lifestyle |
|
|
|
|
|
|
|
|
|
Alcohol consumption |
|
|
|
<0.001 |
|
|
|
<0.001 |
|
None |
3,568,814 (54.2) |
3,490,910 (54.0) |
77,904 (63.2) |
|
865,746 (49.7) |
834,135 (49.3) |
31,611 (63.0) |
|
|
Moderate |
2,358,612 (35.8) |
2,325,044 (36.0) |
33,568 (27.2) |
|
877,689 (50.3) |
859,122 (50.7) |
18,567 (37.0) |
|
|
Excessive |
656,729 (10.0) |
644,991 (10.0) |
11,738 (9.5) |
|
- |
- |
- |
|
|
Tobacco use |
|
|
|
<0.001 |
|
|
|
<0.001 |
|
Never |
4,031,878 (61.2) |
3,951,034 (61.2) |
80,844 (65.6) |
|
852,229 (48.9) |
821482 (48.5) |
30,747 (61.3) |
|
|
Past |
932,941 (14.2) |
912,486 (14.1) |
20,455 (16.6) |
|
337,454 (19.4) |
327832 (19.4) |
9,622 (19.2) |
|
|
Current |
1,619,336 (24.6) |
1,597,425 (24.7) |
21,911 (17.8) |
|
553,752 (31.8) |
543943 (32.1) |
9,809 (19.5) |
|
|
Exercise frequency |
|
|
|
<0.001 |
|
|
|
<0.001 |
|
≥3/week |
3,247,446 (49.3) |
3,184,078 (49.3) |
63,368 (51.4) |
|
831,320 (47.7) |
806,349 (47.6) |
24,971 (49.8) |
|
|
1–2/week |
1,703,632 (25.9) |
1,678,368 (26.0) |
25,264 (20.5) |
|
485,119 (27.8) |
474,002 (28.0) |
11,117 (22.2) |
|
|
None |
1,633,077 (24.8) |
1,598,499 (24.7) |
34,578 (28.1) |
|
426,996 (24.5) |
412,906 (24.4) |
14,090 (28.1) |
|
Table 2.Incidence of hepatocellular carcinoma
Table 2.
|
HCC (n) |
Incidence rate (95% Cl)*
|
Before IPTW |
After IPTW |
After PSM |
|
Univariable analysis |
Multivariable model 1†
|
Multivariable model 2‡
|
Univariable analysis |
Multivariable model 1†
|
Multivariable model 2§
|
Univariable analysis |
|
SHR (95% Cl) |
P-value |
ASHR (95% Cl) |
P-value |
ASHR (95% Cl) |
P-value |
ASHR (95% Cl) |
P-value |
ASHR (95% Cl) |
P-value |
ASHR (95% Cl) |
P-value |
ASHR (95% Cl) |
P-value |
|
Overall |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
No aspirin use |
14,749 |
31.4 (30.9-32.0) |
Reference |
|
Reference |
|
Reference |
|
Reference |
|
Reference |
|
Reference |
|
Reference |
|
|
Aspirin use |
547 |
62.3 (57.2-67.7) |
1.92 (1.76-2.09) |
<0.001 |
0.97 (0.89-1.06) |
0.549 |
0.91 (0.84-1.00) |
0.040 |
0.66 (0.50-0.87) |
0.003 |
0.64 (0.48-0.84) |
0.002 |
0.64 (0.48-0.84) |
0.001 |
0.86 (0.78-0.95) |
0.002 |
|
MASLD |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
No aspirin use |
5,134 |
41.8 (40.7-43.0) |
Reference |
|
Reference |
|
Reference |
|
Reference |
|
Reference |
|
Reference |
|
Reference |
|
|
Aspirin use |
257 |
71.5 (63.0-80.8) |
1.67 (1.47-1.89) |
<0.001 |
0.95 (0.84-1.08) |
0.463 |
0.91 (0.80-1.04) |
0.156 |
0.77 (0.57-1.04) |
0.090 |
0.76 (0.56-1.02) |
0.067 |
0.77 (0.57-1.04) |
0.090 |
0.86 (0.75-0.99) |
0.036 |
Abbreviations
adjusted subdistribution hazard ratio
genome-wide association studies
inverse probability of treatment weighting
metabolic dysfunction-associated steatohepatitis
metabolic dysfunction-associated steatotic liver disease
National Health Insurance Service
propensity score matching
randomized controlled trials
subdistribution hazard ratios
Single nucleotide polymorphisms
REFERENCES
- 1. Rinella ME, Lazarus JV, Ratziu V, Francque SM, Sanyal AJ, Kanwal F, et al. A multisociety Delphi consensus statement on new fatty liver disease nomenclature. J Hepatol 2023;79:1542-1556.
- 2. Yoon EL, Jun DW. Waiting for the changes after the adoption of steatotic liver disease. Clin Mol Hepatol 2023;29:844-850.
- 3. Quek J, Chan KE, Wong ZY, Tan C, Tan B, Lim WH, et al. Global prevalence of non-alcoholic fatty liver disease and nonalcoholic steatohepatitis in the overweight and obese population: a systematic review and meta-analysis. Lancet Gastroenterol Hepatol 2023;8:20-30.
- 4. Le MH, Le DM, Baez TC, Dang H, Nguyen VH, Lee K, et al. Global incidence of adverse clinical events in non-alcoholic fatty liver disease: a systematic review and meta-analysis. Clin Mol Hepatol 2024;30:235-246.
- 5. Huang DQ, El-Serag HB, Loomba R. Global epidemiology of NAFLD-related HCC: trends, predictions, risk factors and prevention. Nat Rev Gastroenterol Hepatol 2021;18:223-238.
- 6. Lin CL, Kao JH. Development of hepatocellular carcinoma in treated and untreated patients with chronic hepatitis B virus infection. Clin Mol Hepatol 2023;29:605-622.
- 7. Harrison SA, Bedossa P, Guy CD, Schattenberg JM, Loomba R, Taub R, et al. A phase 3, randomized, controlled trial of resmetirom in NASH with liver fibrosis. N Engl J Med 2024;390:497-509.
- 8. Loomba R, Bedossa P, Grimmer K, Kemble G, Bruno Martins E, McCulloch W, et al. Denifanstat for the treatment of metabolic dysfunction-associated steatohepatitis: a multicentre, double-blind, randomised, placebo-controlled, phase 2b trial. Lancet Gastroenterol Hepatol 2024;9:1090-1100.
- 9. Hwang IC, Chang J, Kim K, Park SM. Aspirin use and risk of hepatocellular carcinoma in a national cohort study of Korean adults. Sci Rep 2018;8:4968.
- 10. Simon TG, Ma Y, Ludvigsson JF, Chong DQ, Giovannucci EL, Fuchs CS, et al. Association between aspirin use and risk of hepatocellular carcinoma. JAMA Oncol 2018;4:1683-1690.
- 11. Lee TY, Hsu YC, Tseng HC, Yu SH, Lin JT, Wu MS, et al. Association of daily aspirin therapy with risk of hepatocellular carcinoma in patients with chronic hepatitis B. JAMA Intern Med 2019;179:633-640.
- 12. Simon TG, Duberg AS, Aleman S, Chung RT, Chan AT, Ludvigsson JF. Association of aspirin with hepatocellular carcinoma and liver-related mortality. N Engl J Med 2020;382:1018-1028.
- 13. Goh MJ, Sinn DH. Statin and aspirin for chemoprevention of hepatocellular carcinoma: time to use or wait further? Clin Mol Hepatol 2022;28:380-395.
- 14. Choi WM, Kim HJ, Jo AJ, Choi SH, Han S, Ko MJ, et al. Association of aspirin and statin use with the risk of liver cancer in chronic hepatitis B: a nationwide population-based study. Liver Int 2021;41:2777-2785.
- 15. Abdelmalak J, Tan N, Con D, Eslick G, Majeed A, Kemp W, et al. The effect of aspirin use on incident hepatocellular carcinoma-an updated systematic review and meta-analysis. Cancers (Basel) 2023;15.
- 16. Wang S, Zuo L, Lin Z, Yang Z, Chen R, Xu Y. The relationship between aspirin consumption and hepatocellular carcinoma: a systematic review and meta-analysis. Eur J Med Res 2023;28:226.
- 17. Zeng RW, Yong JN, Tan DJH, Fu CE, Lim WH, Xiao J, et al. Meta-analysis: chemoprevention of hepatocellular carcinoma with statins, aspirin and metformin. Aliment Pharmacol Ther 2023;57:600-609.
- 18. Lee TY, Hsu YC, Ho HJ, Lin JT, Chen YJ, Wu CY. Daily aspirin associated with a reduced risk of hepatocellular carcinoma in patients with non-alcoholic fatty liver disease: a population-based cohort study. EClinicalMedicine 2023;61:102065.
- 19. Jang H, Lee YB, Moon H, Chung JW, Nam JY, Cho EJ, et al. Aspirin use and risk of hepatocellular carcinoma in patients with chronic hepatitis B with or without cirrhosis. Hepatology 2022;76:492-501.
- 20. Malehmir M, Pfister D, Gallage S, Szydlowska M, Inverso D, Kotsiliti E, et al. Platelet GPIbα is a mediator and potential interventional target for NASH and subsequent liver cancer. Nat Med 2019;25:641-655.
- 21. Simon TG, Wilechansky RM, Stoyanova S, Grossman A, Dichtel LE, Lauer GM, et al. Aspirin for metabolic dysfunction-associated steatotic liver disease without cirrhosis: a randomized clinical trial. JAMA 2024;331:920-929.
- 22. Kim MN, Park GU, You SC, Lee JS, Lee HW, Kim BK, et al. Aspirin use and risk of HCC and gastrointestinal bleeding in patients with HBV-related cirrhosis: a landmark analysis. J Gastroenterol Hepatol 2025;40:2750-2757.
- 23. Rothwell PM, Price JF, Fowkes FG, Zanchetti A, Roncaglioni MC, Tognoni G, et al. Short-term effects of daily aspirin on cancer incidence, mortality, and non-vascular death: analysis of the time course of risks and benefits in 51 randomised controlled trials. Lancet 2012;379:1602-1612.
- 24. Day N, Oakes S, Luben R, Khaw KT, Bingham S, Welch A, et al. EPIC-Norfolk: study design and characteristics of the cohort. European Prospective Investigation of Cancer. Br J Cancer 1999;80 Suppl 1:95-103.
- 25. Di Angelantonio E, Thompson SG, Kaptoge S, Moore C, Walker M, Armitage J, et al. Efficiency and safety of varying the frequency of whole blood donation (INTERVAL): a randomised trial of 45 000 donors. Lancet 2017;390:2360-2371.
- 26. van der Sluis R. Analyses of the genetic diversity and protein expression variation of the acyl: CoA medium-chain ligases, ACSM2A and ACSM2B. Mol Genet Genomics 2018;293:1279-1292.
- 27. Diggs LP, Greten TF. The effects of platelet accumulation in fatty liver disease. Nat Rev Gastroenterol Hepatol 2019;16:393-394.
- 28. Guidotti LG, Inverso D, Sironi L, Di Lucia P, Fioravanti J, Ganzer L, et al. Immunosurveillance of the liver by intravascular effector CD8(+) T cells. Cell 2015;161:486-500.
- 29. Wang T, Fu X, Jin T, Zhang L, Liu B, Wu Y, et al. Aspirin targets P4HA2 through inhibiting NF-κB and LMCD1-AS1/let-7g to inhibit tumour growth and collagen deposition in hepatocellular carcinoma. EBioMedicine 2019;45:168-180.
- 30. Yoshida S, Ikenaga N, Liu SB, Peng ZW, Chung J, Sverdlov DY, et al. Extrahepatic platelet-derived growth factor-β, delivered by platelets, promotes activation of hepatic stellate cells and biliary fibrosis in mice. Gastroenterology 2014;147:1378-1392.
- 31. Gill D, Georgakis MK, Walker VM, Schmidt AF, Gkatzionis A, Freitag DF, et al. Mendelian randomization for studying the effects of perturbing drug targets. Wellcome Open Res 2021;6:16.
- 32. Chung SW, Moon HS, Shin H, Han H, Park S, Cho H, et al. Inhibition of sodium-glucose cotransporter-2 and liver-related complications in individuals with diabetes: a Mendelian randomization and population-based cohort study. Hepatology 2024;80:633-648.
- 33. Chen L, Rao Q, Gao M, Lv G, Tacke F. Prospects of Mendelian randomization in hepatology: a comprehensive literature review with practice guidance. Clin Mol Hepatol 2025;31:1115-1138.
- 34. Spadafranca A, Rinelli S, Riva A, Morazzoni P, Magni P, Bertoli S, et al. Phaseolus vulgaris extract affects glycometabolic and appetite control in healthy human subjects. Br J Nutr 2013;109:1789-1795.
- 35. Blacklock CJ, Lawrence JR, Wiles D, Malcolm EA, Gibson IH, Kelly CJ, et al. Salicylic acid in the serum of subjects not taking aspirin. Comparison of salicylic acid concentrations in the serum of vegetarians, non-vegetarians, and patients taking low dose aspirin. J Clin Pathol 2001;54:553-555.
- 36. Patel KM, Zhang J, Marsden J, Bays C, Mauldin PD, Schreiner AD. Missed and delayed diagnoses of chronic liver disease in primary care patients with cirrhosis. Dig Dis Sci 2024;69:3721-3728.
- 37. Eslam M, Newsome PN, Sarin SK, Anstee QM, Targher G, Romero-Gomez M, et al. A new definition for metabolic dysfunction-associated fatty liver disease: an international expert consensus statement. J Hepatol 2020;73:202-209.
- 38. European Association for the Study of the Liver (EASL) 1; European Association for the Study of Diabetes (EASD); European Association for the Study of Obesity (EASO). EASL-EASDEASO Clinical Practice Guidelines for the management of nonalcoholic fatty liver disease. J Hepatol 2016;64:1388-1402.
- 39. Park MK, Hur MH, Moon HS, Shin H, Chung SW, Won S, et al. Extrahepatic malignancies in metabolic dysfunction-associated fatty liver disease: a nationwide cohort study. Liver Int 2024;44:799-810.
- 40. Kim MH, Park SM, Chang J, Hwang IC. Association between aspirin and cholangiocarcinoma in a large asian cohort. JAMA Oncol 2020;6:1974-1976.
