Introduction
Metabolic-associated fatty liver disease (MAFLD) is recognized as a leading cause of liver-related morbidity and mortality.1,2 In China, the MAFLD burden is increasing, with prevalence rising from 18% to 29% in the last decade.3 MAFLD comprises a spectrum of disease, ranging from simple steatosis or metabolic-associated fatty liver (MAFL) to the presence of steatohepatitis with varying degrees of fibrosis and cirrhosis.4 MAFLD arises from “multiple hits”, with genes acting as important modifiers of the clinical phenotype.5 Our understanding of the underpinnings of MAFLD has been enhanced by numerous genetic association studies, and all of the polymorphisms identified to date explain only 10–20% of disease heritability.6,7
It is broadly acknowledged that there is overrepresentation of subjects of European ancestry in human genetics research, with ∼79% of all genome-wide association studies (GWAS) participants being of European descent. This overrepresentation hinders a complete understanding of the human genetic architecture. Moreover, it can also have a negative impact, including prediction accuracies between 1.6-4.9-fold lower for other ethnicities than Europeans.8 Hence, increasing the representation of diverse populations and studying other ethnicities has become a research priority.
Several variants in the hydroxysteroid 17-beta dehydrogenase 13 (HSD17B13) gene encoding a hepatic lipid droplet protein have been identified to impact the histological features of MAFLD. However, the impact of HSD17B13 gene variants on MAFLD histology among those of Chinese ancestry is unknown. Notably, allele frequencies, haplotype patterns and the effect size of polymorphisms vary considerably across populations and ethnicities.6 As HSD17B13 has been proposed as a therapeutic target for MAFLD, it is pivotal to explore whether the effect of this variant observed in Caucasian populations extends to other populations, as also to the effect size.
It is known that the genetic association of variants in HSD17B13 with the histological features of MAFLD is complex, with different potentially causative single nucleotide polymorphisms (SNPs) and various SNPs associated with different phenotypic patterns. For example, alleles of rs6834314 and rs72613567 associate with decreased injury and with increased hepatic fat.9 However, there are other studies that show no association of rs72613567 with steatosis.10,11 Noncoding SNPs (e.g., rs6531975) not in linkage disequilibrium with rs72613567 have also been associated with decreased hepatic fat.9 Adding to this complexity, a recent study of 487 patients suggested that those harboring the ‘protective’ TA-allele of rs72613567 have a numerically increased risk for mortality, liver-related death and hepatic decompensation.12 Likewise, while some reports have suggested that there is a potential interaction between HSD17B13 and variants in the patatin-like phospholipase domain containing protein 3 (PNPLA3) gene in MAFLD, subsequent reports have cited a failure to discern an association.13,14
Given these controversies, the aims of this study were 1) to explore the role of variants in the HSD17B13 gene in a cohort of Han Chinese with biopsy-confirmed MAFLD, 2) to clarify the role of the variants on the various morphological features of MAFLD, and 3) to discern if there is any interaction between the variants and variants in PNPLA3.
Methods
Study population
We recruited 427 consecutive Han Chinese patients with biopsy-confirmed MAFLD from the PERSONS cohort (2017.01–2019.05). The definition of MAFLD was based on the criteria proposed by an international expert panel.15 The study cohort included patients from a previously published study as well as additional subjects.16 To ascertain the effects of the HSD17B13 variant on liver disease solely due to MAFLD, patients with other causes of liver disease (including alcohol use disorder or viral hepatitis) were excluded. Briefly, all consecutive patients, aged ≥18, with biopsy-proven MAFLD, and without alternative causes of liver disease were recruited to the study.
The study protocol was approved by the ethics committee of the First Affiliated Hospital of Wenzhou Medical University (2016-246, 1 December 2016) and registered in the Chinese Clinical Trial Registry (ChiCTR-EOC-17013562). Written informed consent was obtained from each subject before their participation in the study. Patient identifiers were anonymized and replaced by the health examination number.
Clinical and biochemical data
Clinical and biochemical data were collected from all patients within 24 hours of liver biopsy. Body mass index (BMI) was calculated as weight (kg) divided by the square of height (m). Insulin resistance (IR) was estimated according by the homoeostasis model assessment (commonly referred to as HOMA).17 Diagnosis of diabetes was based on criteria of the American Diabetes Association.18
Assessment of liver histology
Liver biopsies were performed using a 16-gauge needle under ultrasound guidance. The histology was reviewed by a single liver pathologist (X.D. Wang) who was blinded to the clinical and biochemical data. Histologic scoring was based on the Activity Score.19 Steatohepatitis was diagnosed as a score ≥4 and a score of at least one for each feature of steatosis, ballooning, and lobular inflammation. Severe steatosis, severe ballooning and severe lobular inflammation were defined if their scores were ≥2.
Genetic analysis
Genotyping for the HSD17B13 (rs72613567 and rs6531975) and PNPLA3 (rs738409) variants were performed using the MassARRAY (Agena Biosciences, San Diego, CA, USA) or TaqMan assay (Bio-Rad, Hercules, CA, USA) platforms, according to the manufacturer’s protocol. For the purpose of genotyping, each sample used approximately 20 ng of genomic DNA. Locus-specific PCR and detection primers were designed using Assay Design Suite v3.1.
Statistical analysis
Statistical analyses were performed using R software (v3.5.2; R Foundation for Statistical Computing, Vienna, Austria) and SPSS 19.0 (SPSS Inc., Armonk, NY, USA). Continuous variables were expressed as mean±standard deviation and compared using the one-way analysis of variance test. Categorical variables were expressed as frequency (%) and compared using the chi-square test. The Hardy-Weinberg equilibrium was assessed using the chi-square test. Multivariate logistic regression models were undertaken to test the association between the aforementioned SNPs and liver histology features. A p-value <0.05 was considered to be statistically significant.
Results
Patient characteristics
The study comprised 427 consecutive biopsy-confirmed MAFLD patients; their clinical, biochemical, and histological features are depicted in Supplementary Table 1. The average age was 41 years, with 73.8% being male. About 287 (67.2%) had fibrosis (≥F1), 226 (52.9%) had severe steatosis (S2-S3), 157 (36.8%) had severe ballooning (B2) and 84 (19.7%) had severe inflammation (A2-A3).
Table 1Baseline characteristics of biopsy-confirmed MAFLD patients according to rs72613567 genotypes
| T/T (n=198) | T/TA (n=176) | TA/TA (n=45) | p-value |
|---|
| Age in years | 40.2±11.9 | 41.4±11.5 | 43.1±14.8 | 0.299 |
| Male sex, % | 150 (75.8%) | 126 (71.6%) | 33 (73.3%) | 0.657 |
| Diabetes, % | 63 (31.8%) | 54 (30.7%) | 18 (40.0%) | 0.484 |
| Hypertension, % | 74 (37.4%) | 59 (33.5%) | 22 (48.9%) | 0.161 |
| Waist circumference in cm | 92.2±9.0 | 90.6±8.7 | 91.7±6.8 | 0.212 |
| BMI in kg/m2 | 27.0±3.5 | 26.5±3.3 | 26.3±2.9 | 0.255 |
| HOMA-IR score | 5.3±8.4 | 5.1±6.6 | 6.5±7.5 | 0.541 |
| Platelet count as 109/L | 242.2±61.0 | 246.7±56.2 | 253.1±84.6 | 0.520 |
| Hemoglobin A1c, % | 6.0±1.3 | 6.2±1.5 | 6.3±1.5 | 0.427 |
| Fasting glucose in mmol/L | 5.7±1.5 | 5.5±1.2 | 6.2±2.4 | 0.012 |
| Total cholesterol in mmol/L | 5.2±1.3 | 4.9±1.1 | 5.0±1.0 | 0.100 |
| Triglycerides in mmol/L | 2.4±1.7 | 2.0±1.1 | 2.3±1.3 | 0.044 |
| HDL-cholesterol in mmol/L | 1.0±0.2 | 1.0±0.2 | 1.1±0.4 | 0.019 |
| LDL-cholesterol in mmol/L | 3.1±1.0 | 3.0±0.9 | 2.9±0.8 | 0.331 |
| Albumin in g/L | 46.4±4.2 | 46.4±3.4 | 46.2±3.6 | 0.957 |
| ALT in U/L | 83.4±79.9 | 67.9±56.9 | 70.6±46.6 | 0.079 |
| AST in U/L | 50.4±35.7 | 45.2±35.0 | 40.8±20.6 | 0.139 |
| GGT in U/L | 75.8±83.7 | 68.7±108.9 | 84.6±98.2 | 0.567 |
| Creatinine in µmol/L | 67.1±14.3 | 66.1±12.9 | 70.6±17.4 | 0.159 |
| Uric acid in µmol/L | 395.7±102.9 | 385.8±108.1 | 398.2±120.3 | 0.615 |
| PNPLA3 rs738409 | | | | 0.256 |
| C/C | 56 (28.7%) | 51 (29.7%) | 16 (35.6%) | |
| C/G | 101 (51.8%) | 73 (42.4%) | 19 (42.2%) | |
| G/G | 38 (19.5%) | 48 (27.9%) | 10 (22.2%) | |
Genotype distribution, Hardy-Weinberg equilibrium calculations
Two SNPs in HSD17B13 were genotyped: rs72613567 and rs6531975. The genotype distributions of rs72613567 and rs6531975 in HSD17B13 were in Hardy-Weinberg equilibrium (all, p>0.05). The minor allele frequency (MAF) for rs72613567 and rs6531975 was 0.32 and 0.30 in our cohort, respectively. Each of these MAFs is close to the MAF in general East Asian population in the 1000 Genomes Project.20 The overall genotype distribution of rs72613567 T/T, T/TA and TA/TA was 47.3%, 42.0% and 10.7%, while the distribution of rs6531975 G/G, G/A and A/A was 49.8%, 40.5% and 9.8%, respectively.
Clinical and laboratory characteristics stratified by HSD17B13 variants
The baseline characteristics of study participants according to rs72613567 genotypes is presented in Table 1. There were significant differences in levels of fasting glucose, triglycerides and high-density lipoprotein cholesterol among rs72613567 genotypes (all, p<0.05). Table 2 shows the baseline characteristics of study participants according to rs6531975 genotypes. No significant differences were observed among the rs6531975 genotypes.
Table 2Baseline characteristics of biopsy-confirmed MAFLD patients according to rs6531975 genotypes
| G/G (n=209) | G/A (n=170) | A/A (n=41) | p-value |
|---|
| Age in years | 41.8±12.3 | 40.6±11.2 | 38.9±13.8 | 0.300 |
| Male sex, % | 160 (76.6%) | 122 (71.8%) | 27 (65.9%) | 0.287 |
| Diabetes, % | 61 (29.2%) | 60 (35.3%) | 12 (29.3%) | 0.420 |
| Hypertension, % | 74 (35.4%) | 67 (39.4%) | 14 (34.1%) | 0.672 |
| Waist circumference in cm | 91.6±7.9 | 91.2±9.3 | 90.8±9.8 | 0.824 |
| BMI in kg/m2 | 26.5±3.1 | 26.8±3.6 | 26.7±3.5 | 0.690 |
| HOMA-IR score | 5.8±8.0 | 5.2±8.8 | 4.3±3.5 | 0.472 |
| Platelet count as 109/L | 246.0±62.3 | 243.9±60.9 | 257.4±65.1 | 0.457 |
| Hemoglobin A1c, % | 6.1±1.4 | 6.1±1.4 | 5.9±1.3 | 0.537 |
| Fasting glucose in mmol/L | 5.7±1.6 | 5.7±1.5 | 5.4±1.1 | 0.440 |
| Total cholesterol in mmol/L | 5.0±1.1 | 5.1±1.1 | 5.3±1.6 | 0.324 |
| Triglycerides in mmol/L | 2.2±1.4 | 2.4±1.6 | 2.1±1.0 | 0.284 |
| HDL-cholesterol in mmol/L | 1.0±0.2 | 1.0±0.2 | 1.0±0.2 | 0.665 |
| LDL-cholesterol in mmol/L | 3.0±0.9 | 3.0±0.9 | 3.4±1.2 | 0.061 |
| Albumin in g/L | 46.1±3.6 | 46.5±4.3 | 46.7±3.1 | 0.412 |
| ALT in U/L | 70.3±53.4 | 81.2±93.1 | 84.3±73.5 | 0.275 |
| AST in U/L | 44.1±30.1 | 50.2±40.8 | 51.0±35.7 | 0.193 |
| GGT in U/L | 72.6±103.3 | 76.7±96.9 | 60.9±41.7 | 0.636 |
| Creatinine in µmol/L | 68.0±13.0 | 66.4±15.2 | 63.5±13.7 | 0.137 |
| Uric acid in µmol/L | 390.8±100.9 | 391.6±112.9 | 412.2±115.7 | 0.489 |
| PNPLA3 rs738409 | | | | 0.684 |
| C/C | 62 (30.1%) | 48 (29.1%) | 14 (34.1%) | |
| C/G | 93 (45.1%) | 83 (50.3%) | 16 (39.0%) | |
| G/G | 51 (24.8%) | 34 (20.6%) | 11 (26.8%) | |
HSD17B13 variants and hepatic steatosis
The proportion of severe steatosis in rs72613567 T/T, T/TA and TA/TA was 103 (52.0%), 91 (51.7%)and 27 (60.0%) respectively, while the proportion of severe steatosis in rs6531975 G/G, G/A and A/A was 113 (54.1%), 84 (49.4%) and 24 (58.5%) respectively (Table 3). No association between HSD17B13 variants and severe steatosis was observed in multivariate logistic regression model (Table 4).
Table 3Liver histology features of biopsy-confirmed MAFLD patients according to HSD17B13 genotypes
| HSD17B13 rs72613567
| HSD17B13 rs6531975
|
|---|
| T/T (n=198) | T/TA (n=176) | TA/TA (n=45) | p-value | G/G (n=209) | G/A (n=170) | A/A (n=41) | p-value |
|---|
| Steatosis, n (%) | | | | 0.586 | | | | 0.484 |
| Mild steatosis: <2 | 95 (48.0%) | 85 (48.3%) | 18 (40.0%) | | 96 (45.9%) | 86 (50.6%) | 17 (41.5%) | |
| Severe steatosis: ≥2 | 103 (52.0%) | 91 (51.7%) | 27 (60.0%) | | 113 (54.1%) | 84 (49.4%) | 24 (58.5%) | |
| Hepatocyte ballooning, n (%) | | | 0.226 | | | | 0.401 |
| Mild ballooning: <2 | 125 (63.1%) | 118 (67.0%) | 24 (53.3%) | | 130 (62.2%) | 107 (62.9%) | 30 (73.2%) | |
| Severe ballooning: =2 | 73 (36.9%) | 58 (33.0%) | 21 (46.7%) | | 79 (37.8%) | 63 (37.1%) | 11 (26.8%) | |
| Lobular inflammation, n (%) | | | 0.386 | | | | 0.939 |
| Mild inflammation: <2 | 163 (82.3%) | 141 (80.1%) | 33 (73.3%) | | 169 (80.9%) | 135 (79.4%) | 33 (80.5%) | |
| Severe inflammation: ≥2 | 35 (17.7%) | 35 (19.9%) | 12 (26.7%) | | 40 (19.1%) | 35 (20.6%) | 8 (19.5%) | |
| Presence of fibrosis, n (%) | 135 (68.2%) | 111 (63.1%) | 38 (84.4%) | 0.023 | 150 (71.8%) | 109 (64.1%) | 23 (56.1%) | 0.082 |
Table 4Association between HSD17B13 variants and liver histology features in Chinese MAFLD patients
| SNP | Severe steatosis
| Severe ballooning
| Severe inflammation
| Presence of fibrosis
|
|---|
| OR | 95% CI | p | OR | 95% CI | p | OR | 95% CI | p | OR | 95% CI | p |
|---|
| HSD17B13 rs72613567† |
| Additive model | | | | | | | | | | | | |
| T/T | ref. | – | – | ref. | – | – | ref. | – | – | ref. | – | – |
| T/TA | 1.24 | 0.78–1.96 | 0.368 | 0.93 | 0.60–1.44 | 0.737 | 1.24 | 0.72–2.16 | 0.437 | 0.77 | 0.49–1.20 | 0.252 |
| TA/TA | 1.62 | 0.77–3.42 | 0.203 | 1.37 | 0.69–2.72 | 0.368 | 1.99 | 0.89–4.43 | 0.092 | 2.93 | 1.20–7.17 | 0.019 |
| Dominant model | | | | | | | | | | | | |
| T/T | ref. | – | – | ref. | – | – | ref. | – | – | ref. | – | – |
| T/TA+TA/TA | 1.30 | 0.84–2.02 | 0.234 | 1.01 | 0.67–1.52 | 0.973 | 1.38 | 0.83–2.31 | 0.216 | 0.96 | 0.63–1.48 | 0.867 |
| Recessive model | | | | | | | | | | | | |
| T/T+T/TA | ref. | – | – | ref. | – | – | ref. | – | – | ref. | – | – |
| TA/TA | 1.46 | 0.72–2.98 | 0.292 | 1.42 | 0.74–2.73 | 0.295 | 1.80 | 0.85–3.83 | 0.127 | 3.32 | 1.39–7.91 | 0.007 |
| HSD17B13 rs6531975‡ |
| Additive model | | | | | | | | | | | | |
| G/G | ref. | – | – | ref. | – | – | ref. | – | – | ref. | – | – |
| G/A | 0.69 | 0.44–1.08 | 0.104 | 0.95 | 0.62–1.45 | 0.802 | 0.94 | 0.56–1.60 | 0.830 | 0.65 | 0.42–1.02 | 0.063 |
| A/A | 0.91 | 0.43–1.94 | 0.809 | 0.59 | 0.28–1.24 | 0.164 | 0.84 | 0.35–2.00 | 0.690 | 0.48 | 0.24–0.98 | 0.043 |
| Dominant model | | | | | | | | | | | | |
| G/G | ref. | – | – | ref. | – | – | ref. | – | – | ref. | – | – |
| G/A+A/A | 0.73 | 0.48–1.11 | 0.138 | 0.87 | 0.58–1.30 | 0.496 | 0.92 | 0.56–1.52 | 0.751 | 0.62 | 0.40–0.94 | 0.025 |
| Recessive model | | | | | | | | | | | | |
| G/G+G/A | ref. | – | – | ref. | – | – | ref. | – | – | ref. | – | – |
| A/A | 1.08 | 0.52–2.23 | 0.833 | 0.60 | 0.29–1.24 | 0.170 | 0.86 | 0.37–1.98 | 0.726 | 0.59 | 0.30–1.16 | 0.123 |
HSD17B13 variants and hepatocyte ballooning and lobular inflammation
The proportion of severe ballooning in rs72613567 T/T, T/TA and TA/TA was 73 (36.9%), 58 (33.0%)and 21 (46.7%) respectively, while the proportion of severe ballooning in rs6531975 G/G, G/A and A/A was 79 (37.8%), 63 (37.1%) and 11 (26.8%) respectively. The proportion of severe inflammation in rs72613567 T/T, T/TA and TA/TA was 35 (17.7%), 35 (19.9%) and 12 (26.7%) respectively, while the proportion of severe inflammation in rs6531975 G/G, G/A and A/A was 40 (19.1%), 35 (20.6%) and 8 (19.5%) respectively (Table 3). Both severe ballooning and inflammation were unrelated to HSD17B13 variants in multivariate analysis (Table 4).
HSD17B13 variants and fibrosis
The prevalence of having fibrosis in rs72613567 T/T, T/TA and TA/TA was 135 (68.2%), 111 (63.1%) and 38 (84.4%) respectively. A higher prevalence of fibrosis was observed in patients with the TA/TA genotype in rs72613567 (p<0.05) (Table 3). In rs6531975 genotypes, the prevalence of having fibrosis in G/G, G/A and A/A was 150 (71.8%), 109 (64.1%) and 23 (56.1%) respectively. The A allele carriers of rs6531975 showed a nonsignificant trend for a reduced prevalence of having fibrosis (p=0.082) (Table 3).
To further understand the association between HSD17B13 variants and liver histology in Chinese patients with MAFLD, multivariate logistic regression modeling was undertaken. As shown in Table 4, rs72613567 TA/TA increased the risk of fibrosis with an odds ratio (OR) of 2.93 [TA/TA vs. T/T, 95% confidence interval (CI): 1.20–7.17, p=0.019] for the additive model and an OR of 3.32 (TA/TA vs. T/T+T/TA, 95% CI: 1.39–7.91, p=0.007) for the recessive model after adjusting for age, sex, BMI, presence of diabetes, fasting glucose, triglycerides and high-density lipoprotein cholesterol. In contrast, the rs6531975 A allele appeared to have a protective impact on fibrosis, with an OR of 0.48 (A/A vs. G/G, 95% CI: 0.24–0.98, p=0.043) for the additive model and an OR of 0.62 (G/A+A/A vs. G/G, 95% CI: 0.40–0.94, p=0.025) for the dominant model after adjusting for age, sex, BMI and presence of diabetes.
Interaction of PNPLA3 and HSD17B13 variants
Next, we conducted interaction analysis for HSD17B13 (rs72613567 and rs6531975) and PNPLA3 (rs738409) variants for their impact on liver histology. For fibrosis, no interaction effects were observed between the two genes. In contrast, there was an interaction between rs6531975 and rs738409 with regard to hepatic steatosis (Fig. 1). For the rs738409 risk allele carriers (CG+GG), the proportion of severe steatosis was lower in patients with the rs6531975 A allele (G/A+A/A) compared to those with rs6531975 G/G (Fig. 1A). Using the latter as reference, the rs6531975 A allele (G/A+A/A) attenuated the risk effect of the rs738409 G allele (C/G+G/G) on steatosis, with an OR of 0.57 (95% CI: 0.34–0.96, p=0.034) after adjusting for age, sex, BMI and presence of diabetes (Fig. 1B). The interaction between rs72613567 and rs738409 on liver steatosis was also performed (Fig. 2); however, no effect was observed.
Discussion
We characterized the impact of HSD17B13 gene variants on histological features in a cohort of Han Chinese with MAFLD. This study has three key findings. First, we confirmed the HSD17B13 region as a susceptibility locus for MAFLD-related fibrosis but extended these findings toward the identification of an inverse allelic direction of association as compared to that reported in Europeans. Second, the HSD17B13 variants are only associated with fibrosis and not any other histological feature. Third, the HSD17B13 variants modulate the effect of PNPLA3 rs738409 on hepatic steatosis but no other histological features.
The association between HSD17B13 variants and liver histological features seems to be complex, with multiple suggested functional variants. Notably, in our cohort, the minor allele TA of the rs72613567 variant was related to an increased risk of fibrosis, representing an inverse association as compared to the results in European cohorts. Hence, if there is a shared causal variant across European and Chinese populations, it is unlikely to be rs72613567. In this regard, we observed a protective effect in the minor A allele carriers of the HSD17B13 rs6531975 variant, but this is not in strong linkage disequilibrium with rs72613567. Thus, further fine-mapping studies in Han Chinese populations and comparison to other populations would be helpful to identify shared causal variants across different ethnicities.
The differential effect size and allele direction of variants discovered by GWAS between ethnicities is not uncommon. In one Chinese MAFLD cohort, researchers found that the neurocan (known as NCAN) rs2228603 T variant associated with a higher level of high-density lipoprotein,21 while it was positively related to liver steatosis in the USA population.22 Similarly, toll-like receptor 3 (known as TLR3) rs377529023,24 and interferon lambda-3 (known as IFNL3) rs1297986025,26 variants in Chinese hepatocellular carcinoma populations showed opposite effects to those in non-Asian populations. Inconsistent results have also been observed in other Asian populations, such as among Japanese. For example, tolloid-like 1 (known as TLL1) rs1704720027 and MHC class I polypeptide-related chain A (known as MICA) rs259654228 variants were suggested to have protective impacts on fibrosis and hepatocellular carcinoma in Caucasians. The associations were inverse to those of a Japanese cohort.29,30 Besides, there are several MAFLD-related SNPs in Europeans for which there has been no association in Chinese populations.31–33 Along the same line, lower genetic prediction accuracies (between 1.6-4.9-fold lower) were observed in other ethnicities compared to Europeans.8 Hence, increasing the representation of diverse populations and studying other ethnicities has recently become a research priority to enhance understanding of the human genetic architecture and its translational implications.
The ethnic differences in the characteristics of patients with MAFLD might also contribute to the observed differences in the genetic findings. There is growing evidence, for example, that the MAFLD disease course in Asian populations is different to that in Caucasians. As an example, for the same BMI, there is a higher prevalence of MAFLD in Asians. Published reports also indicate that lean MAFLD accounts for 36.9% of cases in China,3 but only 17.3% of the total disease burden in the USA.34 Differences in metabolic adaptation have been reported between lean and non-lean MAFLD patients, suggesting that lean fatty liver disease likely has a distinct pathophysiology.35
Another intriguing aspect of this study is the lack of association found between HSD17B13 variants and other histological features. To date, the nature of the association between the rs72613567 allelic variant and the histological features of MAFLD, particularly steatosis, is unclear. Abul-Husn and colleagues10 suggested a lack of association between the rs72613567 TA variant and steatosis in human liver, consistent with the study of Pirola et al.11 However, a study by Ma et al.9 found a significant association with hepatic steatosis. Similarly, in animal and in vitro studies, inconsistent results have been reported for an effect of HSD17B13 on hepatic lipid accumulation. Abul-Husn et al.10 showed no differences in lipid accumulation according HSD17B13 isoforms. Similarly Ma et al.9 reported that HSD17B13 overexpression or knockout in HepG2 cells did not affect lipid content. On the other hand, Marion et al.36 noted hepatic steatosis in HSD17B13 knockout mice, whilst Su et al.37 observed steatosis in mice that overexpressed HSD17B13. Collectively, these results imply that HSD17B13 variants could have a direct impact on fibrosis rather than effects on steatosis. These findings may be associated with retinol metabolism, since retinol plays a crucial role in the activation and transformation of hepatic stellate cells to matrix secreting myofibroblasts and the development of hepatic fibrosis.38 Since HSD17B13 participates in the rate limiting step of retinol metabolism,9 the mutant in HSD17B13 might conceivably influence the process of fibrosis.
The interaction between HSD17B13 and PNPLA3 variants in MAFLD is also a subject of controversy.14,39 In this work, we noted an interaction between these variants with regard to steatosis, but not with other histological features. As HSD17B13 has been suggested as a potential therapeutic target for MAFLD and considering the growing concerns about the failure of phase 2 and 3 clinical trials in this disease40,41 that was at least partially attributed to clinical heterogeneity, our study highlights the importance of first understanding the functional basis of the various proposed genomic and other targets before therapeutic development.40,42 Collectively, our data support such an approach. The data from HSD17B13-knockout mice, in fact, suggest that HSD17B13 triggers steatosis and inflammation,36 which is opposite to what has been reported in humans.
The present study has limitations. First, the sample size is modest. In case the observed opposite finding is due to the sample size, we performed a post-hoc power analysis. The power calculated for the model was 72%. It is close to but less than 80%. Considering the low proportion of the rs72613567 TA variant in the general population, we think it is acceptable. In addition, lack of a validation cohort from populations in other parts of China or those of Chinese ancestry living outside mainland China is another limitation.
In conclusion, the HSD17B13 rs72613567 variant appears to be a risk variant for hepatic fibrosis in a Han Chinese MAFLD population, with a different direction for allelic association to that seen in Europeans.
Supporting information
Supplementary Table 1
Baseline characteristics of biopsy-confirmed Chinese MAFLD patients.
(DOCX)
Abbreviations
- BMI:
body mass index
- CI:
confidence interval
- GWAS:
genome-wide association studies
- HOMA:
homoeostasis model assessment
- HSD17B13:
hydroxysteroid 17-beta dehydrogenase 13
- IFNL3:
interferon lambda-3
- IR:
insulin resistance
- MAF:
minor allele frequency
- MAFL:
metabolic-associated fatty liver
- MAFLD:
metabolic-associated fatty liver disease
- MICA:
MHC class I polypeptide-related chain A
- NCAN:
neurocan
- OR:
odds ratio
- PNPLA3:
patatin-like phospholipase domain containing protein 3
- SNP:
single nucleotide polymorphism
- TLL1:
tolloid-like 1
- TLR3:
toll-like receptor 3
Declarations
Funding
This work was supported by grants from the National Natural Science Foundation of China (82070588), High Level Creative Talents from Department of Public Health in Zhejiang Province (S2032102600032) and Project of New Century 551 Talent Nurturing in Wenzhou. GT was supported in part by grants from the University School of Medicine of Verona (Verona, Italy). CDB was supported in part by the Southampton NIHR Biomedical Research Centre (IS-BRC-20004), UK. ME and JG were supported by the Robert W. Storr Bequest to the Sydney Medical Foundation, University of Sydney (Sydney, Australia) and the National Health and Medical Research Council of Australia (NHMRC) Program (APP1053206, APP1149976) and Project (APP1107178 and APP1108422) grants.
Conflict of interest
The authors have no conflict of interests related to this publication.
Authors’ contributions
Study concept and design (WYL, ME, JG, MHZ), acquisition of data (HLM, LJT, GL, PWZ), pathology analysis (XDW), drafting of the manuscript (WYL, ME, JG, KIZ, RSR), critical revision of the manuscript (ME, JG, GT, CDB), statistical analysis (WYL, ME, MZL), study supervision (JG, MHZ), guarantor of the article (MHZ).