Analysis of genetic factors associated with fatty liver disease‐related hepatocellular carcinoma

Abstract Aim Single‐nucleotide polymorphisms (SNPs) in PNPLA3 and hydroxysteroid 17‐beta dehydrogenase 13 (HSD17B13) genes are associated with fatty liver disease (FLD) progression and carcinogenesis. In the present study, we evaluated the characteristics of Japanese FLD patients according to HSD17B13 polymorphisms. Methods We enrolled 402 patients who were clinically and pathologically diagnosed with FLD (alcoholic: 63 cases, nonalcoholic: 339 cases) at our hospital in 1990–2018 (228 males; median age: 54.9 [14.6–83.6] years). FLD patients with HSD17B13 A/A (212 cases) and others (A/AA or AA/AA; 190 cases) were compared. Results Compared to patients with HSD17B13 A/A and others, those with the A/A genotype showed increased incidence of hepatocellular carcinoma (HCC) (A/A vs. others; 18.4% vs. 9.5%, p = 0.01), cardiovascular diseases (14.2% vs. 4.2%, p < 0.01), and hypertension (56.6% vs. 47.4%, p = 0.06). In patients without A/A, the HCC incidence was significantly reduced in those with alcohol‐related FLD, fibrosis‐4 index <2.67, and the PNPLA3 CC genotype; however, there was no significant difference in nonalcoholic‐FLD. Patients without HSD17B13 A/A showed severe steatosis (77% vs. 88.6%, p < 0.01). New HCC developed in 11 cases and the 5–year incidence rate of HCC was 3.3% in patients with both PNPLA3 GG/GC and HSD17B13 A/A, which was significantly higher than the rate for those with other SNP profiles (0.6%, p = 0.03). Conclusions Inhibiting HSD17B13 activity may prevent HCC development, particularly in alcohol‐related FLD and low‐risk patients. Therefore, combinations of SNPs and other risk factors can be used for screening FLD–HCC.


| INTRODUCTION
Nonalcoholic fatty liver disease (NAFLD), nonalcoholic steatohepatitis (NASH), and alcohol-related liver disease (ALD) are chronic liver diseases that have become increasingly common causes of hepatocellular carcinoma (HCC) in Japan, likely due to cure of viral hepatitis. 1hese lifestyle-related diseases are associated with the development of HCC, particularly in patients with type 2 diabetes mellitus. 2In a nationwide survey conducted in Japan, the annual incidence of HCC was 0.11%.5][6] In our previous study, we identified an association between the PNPLA3 GG genotype and fibrosis initiation and development in NAFLD patients. 7Additionally, mutations in hydroxysteroid 17-beta dehydrogenase 13 (HSD17B13), which encodes hepatic lipid droplet protein, are involved in various metabolic processes. 8,9Loss-offunction mutations in the human HSD17B13 gene may confer a strong protective effect against liver injury, inflammation, fibrosis, and HCC onset. 9,10n the present study, we investigated the risk of HCC development according to the PNPLA3 and HSD17B13 SNPs in FLD patients.Additionally, we assessed the associations of HCC development with liver fibrosis and metabolic diseases.

| Study participants and design
This observational single-center study enrolled 402 FLD Japanese patients (339 NAFLD and 63 ALD) diagnosed between 1990 and 2018 at Tokyo Women's Medical University Hospital.The patients were not deliberately chosen and it was a bias-free population.NAFLD was diagnosed according to clinical practice guidelines; 298 patients had biopsyconfirmed NAFLD. 11We confirmed the diagnoses of all study participants and excluded patients who did not fulfill the diagnostic criteria.The pathological stage of NAFLD was evaluated using the classification system of Brunt et al. 12 Advanced fibrosis was defined as fibrosis stage ≥3.For ALD, ethanol intake was defined as >60 g/day. 13e diagnosis of diabetes was based on the American Diabetes Association criteria 14 or the use of anti-diabetic agents.Dyslipidemia and hypertension were diagnosed on the basis of the diagnostic criteria for metabolic syndrome 15,16 or medication use.Laboratory data were collected at the time of liver biopsy or hospital admission.
This study was conducted in accordance with the Declaration of Helsinki and the ethical guidelines of the Tokyo Women's Medical University Hospital (Tokyo, Japan).The Institutional Review Board of Tokyo Women's Medical University Hospital approved the study protocol.Informed consent was obtained from the study participants.

| Analysis of genetic polymorphisms
PNPLA3 and HSD17B13 SNPs were evaluated in 387 and 402 patients, respectively.Genomic DNA was extracted from whole blood.The SNPs rs738409 (PNPLA3) and rs72613567 (HSD17B13) were assessed in patients using the MassARRAY® system (Agena Bioscience) with iPLEX chemistry, which combines mass spectrometry and polymerase chain reaction.The primer sequences (5′-3′) used in this study were designed for the specific sites in each gene (rs738409 F: ACGTT GGA TGA GAG AAA GCC GAC TTACCAC and rs738409 R: ACGTT GGA TGT CAC AGG CCT TGG TATGTTC.rs72613567 F: GCTCT ATT GGT GTT TTA GTA TTT GGGTGTT and rs72613567 R: A GAA GT C T GA TAG AT G G AA TAC TT A C CA ATAAGAA).

| Statistical analysis
Data are presented as medians and ranges.We compared the variables between patients with HSD17B13 A/A vs. A/AA or AA/AA.Differences were assessed by the Kruskal-Wallis test, Mann-Whitney U-test, or χ 2 test using SPSS software (IBM Corp.).P-values <0.05 were considered indicative of statistical significance.We performed multivariate analyses of dataset 1 to explore the potential risk factors of HCC, including age, sex, complications of diabetes mellitus, dyslipidemia, hypertension, BMI, pathological fibrosis, inflammation, and steatosis, and SNPs of PNPLA3 and HSD17B13.The multivariate analyses were repeated for dataset 2 to assess the potential risk factors of age, sex, complications of diabetes mellitus, dyslipidemia, hypertension, BMI, FIB-4 index, and SNPs of PNPLA3 and HSD17B13.Furthermore, the risk of new onset of HCC was evaluated by Cox proportional hazard model for dataset 2. The odds ratios (ORs) and 95% confidence intervals (CIs) were calculated for the HCC risk.
Cumulative curves for new-onset HCC were constructed using the Kaplan-Meier method.The incidence of HCC was compared using the log-rank test.

| Demographic characteristics of FLD patients with HSD polymorphism
This study included 402 Japanese patients with FLD who were diagnosed between 1990 and 2018 (63 with ALD and 339 with NAFLD).The diagnosis was confirmed on biopsy in 298 of these patients.We compared the clinical features of FLD patients according to the HSD17B13 polymorphisms (Table 1).Compared to patients with HSD17B13 wild-type A/A (212 cases) and variants (A/AA or AA/AA, 190 cases), there were no significant differences in the age, sex distribution, BMI, ALD, or prevalence of diabetes or dyslipidemia.Patients with HSD17B13 variants had significantly lower rates There were no significant differences in the other laboratory findings, including the FIB-4 index, among patients with various HSD17B13 polymorphisms.There were no significant differences in the percentage of liver tissue with severe fibrosis (F ≥ 3) or severe inflammation (A ≥ 2) among patients with various HSD17B13 polymorphisms.Severe steatosis was more common among patients with HSD17B13 variants than those with wild-type polymorphism (S ≥ 2, 77.3% vs. 88.6%,p < 0.01).

| HCC development according to gene polymorphisms and complications
HCC was common in patients with HSD17B13 A/A and PNPLA3 GG/GC (Figure 1A, p = 0.047).Among patients with PNPLA3 CC, no cases of HCC developed in those with HSD17B13 variants (Figure 1B, p = 0.15).In combination with high-risk SNPs, PNPLA3 GG/GC and HSD17B13 AA were associated with a significantly increased risk of HCC (Figure 1C, p < 0.01).Conversely, there were no HCC F I G U R E 1 HCC complications stratified according to genetic factors.HCC incidence by (A) PNPLA3 GG/GC and (B) PNPLA3 CC according to the HSD17B13 SNP, (C) combinations of risk SNPs, and (D) combinations of non-risk SNPs.HCC was frequently observed in patients with HSD17B13 A/A in PNPLA3 GG/GC (A, p = 0.047).Among cases with PNPLA3 CC, no HCC was observed among those with HSD17B13 variants (B, p = 0.15).The incidence of HCC was significantly increased in patients with PNPLA3 GG/GC and HSD17B13 A/A (C, p < 0.01).Conversely, there were no HCC cases among patients with PNPLA3 CC and HSD17B13 variants (D, p = 0.05).HCC, hepatocellular carcinoma; HSD17B13, 17-beta dehydrogenase 13; PNPLA3, patatin-like phospholipase 3; SNP, single-nucleotide polymorphism.
In patients with HSD17B13 variants, the HCC incidence was significantly reduced in alcohol-related FLD (HSD17B13 variants, OR = 0.29, 95% CI = 0.10-0.85,p = 0.02, Figure 2A).However, the risk of HCC was not increased in NAFLD (p = 0.37, Figure 2B).After stratification according to the FIB-4 index, HCC was more common in HSD17B13 A/A than HSD17B13 variants among those with a FIB-4 index <2.67 (p = 0.04, Figure 2C), but not among those with a FIB-4 index ≥2.67 (OR = 0.37, 95% CI = 0.14-0.98,p = 0.04, Figure 2D).The incidence of HCC was reduced in non-obese patients with HSD17B13 variants (p = 0.04, Figure 2E), but not in obese patients (p = 0.12, Figure 2F).The incidence of HCC was increased in patients with the HSD17B13 A/A genotype among diabetic patients (p = 0.04, Figure 2H).HCC was more common in patients with the HSD17B13 A/A genotype (p = 0.03, Figure 2J) than those with HSD17B13 variants in hypertensive patients.However, there were no significant differences in HCC incidence among patients without complications of diabetes or hypertension (p = 0.12, Figure 2G and p =0.36, Figure 2I, respectively).The incidence of HCC was increased in patients with HSD17B13 A/A in the absence of dyslipidemia (p = 0.03, Figure 2K).Furthermore, there were no significant differences in the risk of HCC development in patients with dyslipidemia stratified according to HSD17B13 polymorphisms (p = 0.12, Figure 2L).

| New-onset HCC development stratified according to PNPLA3 and HSD17B13 polymorphisms
Eleven new cases of HCC developed over 8.1 (0.5-25.1) years of the observation period (n = 370).Among patients with the PNPLA3 CC genotype, none developed HCC in 5 years, whereas HCC was observed in 2.1% of cases with PNPLA3 GG/GC after 5 years (p = 0.17, Figure 3A).HCC developed in 2.2% of patients with HSD17B13 A/A and 0.6% of patients with HSD17B13 variants over 5 years (p = 0.07, Figure 3B).In patients with both PNPLA3 GG/ GC and HSD17B13 A/A, HCC was significantly more common than in patients with other SNPs (p = 0.03, Figure 3C).

| DISCUSSION
This study assessed the rate of HCC development in FLD patients according to the genetic polymorphisms of PNPLA3 and HSD17B13.The genetic background of FLD may be associated with increased or decreased risk of HCC.Previous studies have shown that PNPLA3 GG/GC is significantly associated with an increased risk of HCC development in FLD, 3 whereas HSD17B13 variants are associated with a reduced risk of HCC onset. 10Our cohort study showed similar results; the combination of high-risk alleles of PNPLA3 and HSD17B13 should prompt frequent screening for HCC.Among patients with HSD17B13 variants, the HCC risk was reduced in those with a FIB-4 index <2.67,normal body weight, and normal lipid levels; conversely, the HCC risk was increased in patients with diabetes mellitus and hypertension among those with the HSD17B13 A/A genotype.Liver fibrosis and obesity may be more strong factors for carcinogenesis than suppression by genetic factors.
Gellert-Kristensen et al. 17 demonstrated that HSD17B13 rs72613567: TA is associated with decreased alanine aminotransferase levels in the Danish general population.Although there were no significant abnormalities in liver enzymes in the present study, the serum triglyceride level was likely to be increased in patients with HSD17B13 variants.Rotroff et al. 18 demonstrated that HSD17B13 variants are associated with higher levels of triglycerides and high-density lipoprotein levels in White populations.Our data showed that the incidence rates of CVD and hypertension were significantly in patients with HSD17B13 variants.No evidence is available to confirm whether or not HSD17B13 variants protect against CVD or hypertension.
In terms of the effect of HSD17B13 on liver pathology, the rs72613567:TA variant is associated with a reduced risk of NASH but not steatosis. 9Ma et al. 19 demonstrated that HSD17B13 functions as a retinol dehydrogenase and is associated with histological features of NAFLD.However, it does not affect hepatocyte lipid content directly.In our study, HSD17B13 variants increased the pathological fat content and frequency of S ≥ 2 compared to the HSD17B13 A/A genotype.In Japanese patients with NAFLD, the HSD17B13 rs6834314 G allele was associated with a significantly lower prevalence of severe inflammation and ballooning, as well as a higher prevalence of advanced steatosis. 20In carriers of this allele, the effects of the PNPLA3 rs738409 GG genotype on advanced hepatic fibrosis are attenuated. 20However, the prevalence of advanced fibrosis and severe inflammation was not significantly increased in those with HSD17B13 variants in this study.
Yang et al. 10 demonstrated the protective effects of the HSD17B13 rs72613567: TA allele on HCC development.Among 3315 patients, HCC was reduced in those with ALD F I G U R E 3 New-onset HCC rate according to (A) PNPLA3, (B) HSD17B13, and (C) both PNPLA3 + HSD17B13 SNP status, as determined by Kaplan-Meier curves.There were no significant differences and a trend toward an increased risk of HCC development in patients with the PNPLA3 genotype (A, p = 0.17) and HSD17B13 genotype (B, p = 0.07), respectively.The incidence of de novo HCC was significantly increased in patients with PNPLA3 GG/GC and HSD17B13 A/A (log-rank = 4.93, p = 0.03).HCC, hepatocellular carcinoma; HSD17B13, 17-beta dehydrogenase 13; PNPLA3, patatin-like phospholipase 3; SNP, single-nucleotide polymorphism.
(OR = 0.64). 10Husn et al. 9 showed that the minor allele of HSD17B13 rs72613567 reduced the risk of ALD by 53% and of NAFLD by 30%.Our study also showed that the risk of HCC development was reduced with this allele, particularly in patients with ALD.The incidence of HCC was also reduced in patients with a FIB-4 index <2.67,normal weight, and normal lipid levels.Among patients with diabetes and hypertension, HSD17B13 A/A carriers have a higher risk of HCC.Therefore, HSD17B13 may be an independent protective factor against HCC development, in association with the low-risk factors of fibrosis and obesity.Conversely, among patients with high-risk factors of HCC, such as diabetes and hypertension, the HSD17B13 A/A genotype was associated with an increased risk of HCC.In other words, the incidence of HCC was reduced in patients with HSD17B13 variants who had diabetes and hypertension, although there was no significant difference among patients without diabetes and hypertension.It is possible that our results were due to the fact that the incidence of HCC was not reduced in NAFLD patients with HSD17B13 variants (data not shown).The incidence of HCC development is low in NAFLD patients without diabetes or hypertension and can be affected by the genetic background.
In a previous study, multivariate analysis showed that age, male sex, diabetes, platelet count, gammaglutamyltransferase and albumin levels, and the 7-SNP genetic risk score, including PNPLA3 and HSD17B13, are independent risk factors of HCC development. 21We performed a multivariate analysis of two independent datasets to determine the risk factors of HCC.Dataset 1 included pathological fibrosis, whereas dataset 2 included the FIB-4 index.Although our study did not identify genetic polymorphism as an independent risk factor of FLD-HCC, age, male sex, hypertension, and fibrosis were major risk factors of HCC.Low inflammation and normal lipid level may result from liver cirrhosis.These factors are more strongly related to carcinogenesis in FLD compared to genetic factors.
In the analysis of new-onset HCC, HSD17B13 variants were associated with a decreased 5-year incidence rate of HCC (0.6%, p = 0.07); however, the risk was not increased with PNPLA3.The combination of high-risk alleles of PNPLA3 GG/GC and HSD17B13 A/A was associated with a significantly higher risk of HCC compared to other SNPs (p = 0.03), indicating that it is a good screening tool for HCC development in FLD.
This study had several limitations.First, this was a single-center observational study.Second, the number of new cases of HCC was small, and evaluation of HCC development was mainly in the patients with HCC complications, it was not enough in the patients with new HCC.However, it was not particularly low in carcinogenesis of NAFLD in our institute, in a nationwide survey conducted in Japan, the annual incidence of HCC was 0.11%. 2 The incidence of HCC among patients with non-cirrhotic NAFLD is reported approximately from 0.1 to 1.3 per 1000 patient-years. 22Third, only a small number of ALD patients were included.Fourth, in cirrhosis, synthesis of lipid was decreased and low hemoglobin levels due to pancytopenia or bleeding may mask dyslipidemia or diabetes.Finally, we could not separately assess the influence of therapeutic drugs and lifestyle modification including body weight reduction and nutrition of the patients.Future studies should include larger sample sizes.

T A B L E 1
Characteristics of patients with NAFLD according to HSD17B13 polymorphism.

T A B L E 2 HCC
risk factors according to multivariate analysis.