SAMM50-rs2073082, -rs738491 and -rs3761472 Interactions Enhancement of Susceptibility to Non-Alcoholic Fatty Liver Disease

Background and aim: Several studies have identified that three SAMM50 polymorphisms (rs2073082, rs738491, rs3761472) are associated with an increased risk of non-alcoholic fatty liver disease (NAFLD). However, the clinical significance of the SAMM50 SNP in relation to NAFLD remains largely unknown. Therefore, we conducted a clinical study and SNP–SNP interaction analysis to further elucidate the effect of the SAMM50 SNP on the progression of NAFLD in the elderly. Methods: A total of 1053 patients over the age of 65 years were recruited. Liver fat and fibrosis were detected by abdominal ultrasound or FibroScan, respectively. Genomic DNA was extracted and then genotyped by Fluidigm 96.96 Dynamic Array. Multivariable logistic regression was used to evaluate the association between NAFLD and SNP. SNP–SNP interactions were analyzed using generalized multivariate dimensionality reduction (GMDR). Results: The risk of NAFLD was substantially higher in people who carried SAMM50-rs2073082 G and -rs738491 T alleles (OR, 1.962; 95% CI, 1.448–2.659; p < 0.001; OR, 1.532; 95% CI, 1.246–1.884; p = 0.021, respectively) compared to noncarriers. Carriers of the rs738491 T and rs3761472 G alleles in the cohort showed a significant increase in liver stiffness measurements (LSM). The combination of the three SNPs showed the highest predictive power for NAFLD. The rs2073082 G allele, rs738491 T allele and rs3761472 G carriers had a two-fold higher risk of NAFLD compared to noncarriers. Conclusions: Our research has demonstrated a strong correlation between the genetic polymorphism of SAMM50 and NAFLD in the elderly, which will contribute to a better understanding of the impact of age and genetics on this condition. Additionally, this study provides a potential predictive model for the early clinical warning of NAFLD.


Introduction
Nonalcoholic fatty liver disease (NAFLD) is the leading cause of chronic liver disease, affecting up to 1.7 billion individuals worldwide and posing a significant health burden.Notably, China has one of the highest prevalence, morbidity and annual mortality rates associated with NAFLD in Asia [1,2].NAFLD is caused by a confluence of genetic and environmental factors, with hereditary predisposition accounting for approximately 50% of the risk [3,4].Genome-wide association studies (GWASs) have identified dozens of genetic Biomedicines 2023, 11, 2416 2 of 15 variants related to NAFLD over the last decade, such as patatin-like phospholipase domaincontaining protein 3 (PNPLA3), Glucokinase regulatory protein (GCKR) and Membrane Bound O-Acyltransferase Domain Containing 7 (MBOAT7) [5][6][7].This indicates that SNPs play a vital role in the development of NAFLD.
Recently, the SAMM50 variant was found to be closely associated with NAFLD vulnerability [8][9][10].The SAMM50 gene encodes Sam50, which is a kind of β-barrel protein distributed in the mitochondrial membrane [11].It is also involved in the regulation of mitochondrial morphology, function and scavenging mitochondrial reactive oxygen species (ROS) [12].Several SNPs in the SAMM50 gene have also been reported to be associated with an increased risk of NAFLD.Clinical studies have revealed that SAMM50 variants, including rs738491 and rs3761472, increase susceptibility to NAFLD [12,13].Subsequent reports have also suggested that rs2073082 is associated with an increased risk of NAFLD [10].Furthermore, Kitamoto et al. discovered a significant association between the rs738491 variant and fibrosis in a Japanese cohort, whereas rs3761472 did not show a correlation [12].However, rs738491 was not found to be associated with fibrosis in a Chinese cohort [14].The findings are inconsistent and require further investigation.In short, the clinical significance of the SAMM50 SNP in relation to NAFLD remains largely unknown.
In addition, the trend of global aging is becoming increasingly obvious [15].The World Health Organization (WHO) estimates that the number of individuals aged 60 and above will reach approximately 840 million by 2025 [16].Evidence indicates that aging increases the incidence of NAFLD, especially mortality from nonalcoholic steatohepatitis (NASH) [7,[17][18][19][20].Unfortunately, little attention has been paid to NAFLD in older adults.Therefore, we conducted this case-control study to investigate the effects of three variants of SAMM50 and their interactions on NAFLD in the elderly Chinese population.
In the present study, we evaluated the effects of three variants of SAMM50 and their interactions on the development of NAFLD.The results demonstrated that all three variants were associated with NAFLD in the elderly.Among them, carriers of the rs2073082 G allele and rs738491 T allele were associated with susceptibility to NAFLD, and carriers of rs738491 T and rs3761472 G alleles were associated with fibrosis.In addition, GMDR analysis demonstrated that the best model for predicting NAFLD included all three SNPs.Thus, this study provides important insights into genetic factors, which may contribute to a better understanding of the underlying mechanisms involved.

Patients and Methods
Elderly citizens in the Beijing Mentougou community who participated in annual free physical examinations were recruited from 1 November 2020 to 30 September 2021.The study protocol was approved by the Ethics Committee of Beijing You'an Hospital, Capital Medical University (IRB number (2020)-133).The approval date was 28 October 2020.The registration number was ChiCTR 2100043106.All subjects signed informed consent forms.

Patient Selection and Enrollment Criteria
NAFLD was diagnosed according to the 2018 AASLD NAFLD management guidelines (i.e., defined as evidence of hepatic steatosis on abdominal ultrasound) [21].In addition to a diagnosis of NAFLD, the inclusion criteria included the following reasons: (1) residents in the community; (2) volunteered participating in the annual free physical examination, which was provided by the Beijing government for residents older than 65; (3) signed informed consent.Participants were excluded from analyses for the following reasons: (1) missing data on genetic polymorphisms or other important laboratory parameters or medical history; (2) excessive alcohol consumption (>140 g/week for men and >70 g/week for women); (3) inability to obtain reliable abdominal ultrasound results due to specific reasons, such as intestinal gas interference; (4) malignant tumors, HIV and other serious diseases that may affect the nutritional status or organ function; (5) comorbidity with other liver diseases, such as viral hepatitis and autoimmune hepatitis.Typically, 10-20 residents took part in the routine examination every morning.If a patient was diagnosed with fatty liver by ultrasound, they were transferred to an isolated room where they were recruited and underwent further testing.Controls were selected from residents who arrived after previous NAFLD patients and did not have fatty liver.Ultimately, 1423 residents were examined, and 1053 residents were recruited into the study (Figure 1).
g/week for women); (3) inability to obtain reliable abdominal ultrasound results d specific reasons, such as intestinal gas interference; (4) malignant tumors, HIV and serious diseases that may affect the nutritional status or organ function; (5) comor with other liver diseases, such as viral hepatitis and autoimmune hepatitis.
Typically, 10-20 residents took part in the routine examination every mornin patient was diagnosed with fatty liver by ultrasound, they were transferred to an is room where they were recruited and underwent further testing.Controls were se from residents who arrived after previous NAFLD patients and did not have fatty Ultimately, 1423 residents were examined, and 1053 residents were recruited in study (Figure 1).

Data Collection
Baseline information for all participants, including their demographics, anth metrics, clinical parameters, and comorbidities, was measured.All laboratory tests performed in the central lab of Menkuang Hospital or Beijing Jingmei Group Genera pital, including serum concentrations of alanine aminotransferase (ALT), aspartate notransferase (AST), triglyceride (TG), total cholesterol (TC), high-density lipop (HDL), and low-density lipoprotein (LDL), Blood glucose (GLU) and Glycosylated H globin, Type A1c (HbA1c).Two trained investigators performed an abdominal ultra and determined liver fat content and liver stiffness with the FibroScan 502 touch d (Echosens, Paris, France).

Data Collection
Baseline information for all participants, including their demographics, anthropometrics, clinical parameters, and comorbidities, was measured.All laboratory tests were performed in the central lab of Menkuang Hospital or Beijing Jingmei Group General Hospital, including serum concentrations of alanine aminotransferase (ALT), aspartate aminotransferase (AST), triglyceride (TG), total cholesterol (TC), high-density lipoprotein (HDL), and low-density lipoprotein (LDL), Blood glucose (GLU) and Glycosylated Hemoglobin, Type A1c (HbA1c).Two trained investigators performed an abdominal ultrasound and determined liver fat content and liver stiffness with the FibroScan 502 touch device (Echosens, Paris, France).

Genomic DNA Extracting and Genotyping
Genomic DNA was extracted from the patient's blood specimens by BGI-Shenzhen, China.The concentration and quality of DNA were determined by spectrophotometry (Nanodrop 2000, Thermo Scientific, Wilmington, DE, USA) and diluted to approximately 50 ng/mL before genotyping.Next, genomic DNA was genotyped with the use of a 96.96 genotyping integrated fluidics circuit with customized SNP-type assays on the Juno TM system (Fluidigm, South San Francisco, CA, USA), and quantification on the Biomark TM (Fluidigm, South San Francisco, CA, USA) in accordance with the manufacturers' instructions.The data were analyzed using Fluidigm SNP Genotyping Analysis software version 4.5.1 (South San Francisco, CA, USA).

Statistical Analysis
The unpaired Student's t-test, one-way ANOVA, or Mann-Whitney U-test (if the data were not normally distributed) were employed for comparing continuous variables, which were presented as means with standard deviations (SD) or medians (25th and 75th percentiles).Categorical variables were reported as numbers and percentages and subjected to the χ 2 test or Fisher's exact test.The χ 2 test was utilized to determine the Hardy-Weinberg equilibrium.Using the χ 2 test, differences in alleles and genotype distributions between groups were calculated.By using logistic regression analysis, the relationship between genetic variations and NAFLD was assessed, and the odds ratio (OR) with a 95% confidence interval (CI) was obtained.Open source Java software versions 3.0.2Multifactor Dimension Reduction (MDR) (http://www.epistasis.org/mdr.htmlaccessed on 4 August 2023) [26] and 0.9 Generalized Multivariate Dimensionality Reduction (GMDR) (http://www.ssg.uab.edu/gmdr/accessed on 4 August 2023) [27,28] for analyzing SNP-SNP interactions.The statistical analyses were performed using the Statistical Package for Social Sciences (SPSS), version 26.0 (SPSS Inc., Chicago, IL, USA).A p value < 0.05 was considered statistically significant.

Baseline Characteristics of Non-NAFLD Controls and NAFLD Patients
A total of 1053 subjects were included, and 28.7% of the subjects were male.Among these patients, 590 (56.03%) had NAFLD.The demographic and clinical characteristics of the two groups are listed in Table 1.The NAFLD patients had a higher BMI, waist and hip circumferences, serum levels of ALT, AST, HbA1C, HOMA-IR, insulin and serum TG and lower HDL than the control group.Controlled attenuation parameter (CAP) and liver stiffness measurement (LSM) were also significantly higher in the NAFLD group than in the controls.However, no significant differences were observed between the two groups in terms of TC, LDL and GLU.Regarding the metabolic profiles, among NAFLD cases, 465 subjects (78.81%) had hypertension, 272 subjects (46.10%) had T2DM, 537 subjects (91.02%) had Mets, and 425 subjects (72.03%) were obese.The prevalence showed differences between the NAFLD and non-NAFLD groups (p < 0.05).Meanwhile, there were no significant differences in stroke incidence and lipid-lowering agent use between the two groups.

Genotypes and Allele Frequencies of rs2073082, rs738491 and rs3761472 in Non-NAFLD and NAFLD Groups
The distribution of each SNP in the non-NAFLD and NAFLD groups was consistent with the Hardy-Weinberg balance and was representative of the population (p > 0.05).As described in Table 2, there were significant differences in the genotype of rs2073082 between the NAFLD group and the control group.The genotype and allele frequencies of rs738491 differed between the two groups.However, the distribution of the rs3761472 genotype and allele frequencies was comparable.The relationship between the three variants of SAMM50 and NAFLD susceptibility under different genetic models is analyzed in Table 3.There was a noticeable association between rs2073082 polymorphism and NAFLD susceptibility under the homozygous model (GG vs. AA, OR = 1.638, 95% CI, 1.222-2.196,p < 0.001) and recessive model (AG + GG vs. AA, OR = 1.836, 95% CI, 1.392-2.421,p < 0.001).Significant association remained after adjusting for confounding factors (adjusted OR, 1.691; 95% CI, 1.235-2.315;p < 0.001; adjusted OR, 1.962; 95% CI, 1.448-2.659;p < 0.001, respectively).Carriers of the G-allele had a higher risk of NAFLD.Moreover, the data demonstrated strong links between rs738491 and the presence of NAFLD under the allelic model (T vs. C, adjusted OR, 1.216; 95% CI, 1.005-1.472;p = 0.045), the homozygous model (TT vs. CC, adjusted OR, 1.373; 95% CI, 1.050-1.376;p = 0.021) and the recessive model (CT + TT vs. CC, adjusted OR, 1.532; 95% CI, 1.246-1.884;p = 0.021), which indicated that the T-allele was the risk allele of NAFLD.However, no significant difference existed between any model of rs3761472 and NAFLD.

Association of Three SAMM50 Gene Variants with Clinical Features
Since previous studies have indicated that the G-allele in rs2073082, T-allele in rs738491 and G-allele in rs3761472 all increased susceptibility to NAFLD [10,12,13], we classified the patients as carriers and noncarriers in the whole population.As shown in Table 4, the LSM of the rs738491 T allele and rs3761472 G allele carriers was significantly higher than that of noncarriers (p = 0.040, p = 0.019, respectively).Furthermore, our analysis revealed that the significance of LSM remained even after adjusting for sex, age and BMI (Table S1).As for rs3761472, the TC and LDL levels had significant differences between the two groups, even after adjusting for sex, age, BMI and use of lipid-lowering agents in the logistic regression model (Table S2).Nevertheless, the levels of ALT, AST, TG and HDL between the carriers and the noncarriers of the three genetic variants were not statistically different.The clinical characteristics were also compared in the whole population according to the genotypes of the three SNPs.After correction for multiple linear regression, LSM indicated a difference between groups (Tables S3 and S4).There were no differences between the other indicators.
Similar analyses were also performed in the NAFLD population.No statistical significance was discovered (Tables S5 and S6).

Analysis of SNP-SNP Interactions
GMDR was used to analyze the interactions between these SNPs.Table 5 presents a summary of the results obtained from the GMDR analysis for the one-to three-locus models.Among them, a significant three-locus model (p = 0.0107) involving rs2073082, rs738491 and rs3761472 was identified, demonstrating the highest training accuracy (56.97%), testing accuracy (TA, 55.52%) and the best cross-validation consistency (CVC: 10/10).A graphical presentation of the best identified interaction models in NAFLD and control groups is given in Figure 2, which demonstrated the distribution of high and low risk of NAFLD in a three-locus genetic model combination (Figure 2).The dendrogram (Figure 3) showed that rs2073082 and rs738491 exhibited antagonism effects on NAFLD susceptibility.Fruchterman-Reingold (Figure 4) revealed that rs3761472 synergized with both rs738491 and rs2073082, resulting in positive information gain values of 0.21% and 0.04% in NAFLD, while rs2073082 and rs738491 were antagonistic with negative IG values (−0.76%).Age, sex and BMI adjustments were made to the models.
Importantly, NAFLD risk increased in parallel with the number of loci even after adjusting for confounders.The adjusted OR of one locus, two locus and three locus models were 1.532 (95% CI: 1.144-2.053),1.809 (95% CI: 1.147-2.853)and 1.892 (95%CI: 1.196-2.993),respectively.Additionally, rs2073082 G carriers, rs738491 T carriers and rs3761472 G carriers had a two-fold higher risk of NAFLD compared to noncarriers (Table 6).Interactions between SNPs are depicted as lines in different colors.Synergy is depicted as an orange line between SNPs accompanied by a positive percent of entropy, while redundancy is indicated as a blue line accompanied by a negative percent of entropy.Rs3761472 synergized with both rs738491 and rs2073082, resulting in positive information gain values of 0.12% and 0.04% in NAFLD, while rs2073082 and rs738491 were antagonistic with negative IG values (−0.76%).

Discussion
In the last decade, it has become increasingly clear that genetic markers significantly impact the progression of NAFLD.For example, the variant PNPLA3 rs738409 is associated with increased liver fat content and fibrosis [29].GCKR rs780094 can also cause hepatic steatosis, impair mitochondrial β-oxidation and cause a predisposition to fatty liver-related liver disease [30].Therefore, the identification of genetic markers for NAFLD can be a valuable tool in clinical management, helping with risk assessment and highlighting potential treatment targets [5,31,32].In the current study, we conducted clinical research and analyzed the interaction between the novel genetic factor SAMM50 SNPs (rs2073082, rs738491 and rs3761472) to understand their impact on the progression of NAFLD in elderly patients.The major findings of the study are as follows: (i) The carriers of the rs2073082 G allele and rs738491 T allele significantly increased susceptibility to NAFLD; (ii) the rs738491 T allele and rs3761472 G allele carriers had significantly higher levels of LSM than the noncarriers in the whole population; (iii) We found that the best model for predicting NAFLD in elderly individuals included rs2073082, rs738491 and rs3761472 using GMDR analysis.
Aging, a major risk factor for increased susceptibility to certain diseases, is characterized by a gradual physiological process that affects all systems of the body, including cells, tissues and organs, which ultimately leads to impaired biological function of the organism [33].Various endogenous and exogenous factors, such as genetic makeup or dietary habits, also have a significant impact on individualized aging [34].It has been reported that aging, particularly hepatic age, can promote the development of steatosis, non-alcoholic steatohepatitis (NASH) and hepatocellular carcinoma by reducing the regenerative capacity, biotransformation and increasing inflammation of the liver [34].In a territory-wide retrospective cohort study, Zhang et al. [35] found that most patients with NAFLD and T2DM developed liver-related events or cirrhosis after age 50, suggesting that aging is an independent and strong predictor of NAFLD.Another study, including a cohort of 182 Japanese patients with biopsy-proven NAFLD, also demonstrated that age was closely related to the pathological progression of liver fibrosis [36].However, epidemiological data on fatty liver in the elderly population are still much less available compared to middle-aged adults and children.Based on the fact that the global elderly population continues to increase, research on aging and NAFLD may contribute to reducing the burden on healthcare systems.
The SAMM50 gene and its encoded protein SAM50 are important components of the sorting and assembly machinery complex in the mitochondrial outer membrane, involved in maintaining the stability of mitochondrial DNA, respiratory chain complexes, mitochondrial cristae structure [12,14], and the regulation of mitophagy [37].It has been reported that SAM50 deficiency triggers membrane remodeling and mitochondrial dysfunction, reducing the ability to clear reactive oxygen species (ROS) and causing lipotoxicity and hepatocyte damage, which further contributes to the development of NAFLD [10].Polymorphisms in rs2073082, rs738491 and rs3761472 of the SAMM50 gene were also found to be significantly associated with susceptibility to NAFLD in middle-aged Japanese and Chinese populations through GWAS and a clinical cohort study, respectively [12,13].Individuals with the homozygous TT genotype of rs738491 and the GG genotype of rs2073082 had a lower expression of the SAMM50 gene, but only decreased protein levels of rs738491 TT were found in the livers of patients with NAFLD [10].The above studies imply that these variants may promote mitochondrial dysfunction by reducing SAMM50 expression, which further leads to the development of NAFLD.
Our results indicate that the rs2073082 G allele and rs738491 T allele variations in the SAMM50 gene significantly increase susceptibility to NAFLD in an elderly population, which is consistent with previous studies reported by Zuyin et al. in a cohort of Han Chinese adults [10].Furthermore, another population-based study with an average age of around 41 years showed that the rs3761472 G allele was associated with an increased susceptibility to non-alcoholic fatty liver disease, but this was not found in our cohort.Furthermore, the current research on the relationship between SAMM50 genetic polymorphism and liver fibrosis is inconsistent.A biopsy-proven study based on a cohort of Japanese liver fibrosis patients with a mean age of approximately 50 years demonstrated that the rs738491 gene polymorphism, but not rs3761472, was strongly associated with fibrosis progression, whereas rs738491 was not found to be related to fibrosis in another study of a biopsy-proven Chinese liver fibrosis cohort with a mean age of approximately 40 years [12,14].Unlike the above, our results showed that both the rs738491 T allele and rs3761472 G allele carriers were related to liver fibrosis in the whole aging population but not in the NAFLD group.The conflicting findings between SAMM50 polymorphism and non-alcoholic fatty liver disease may require further validation through large-scale clinical cohorts and in-depth mechanistic studies.However, our finding of this unique correlation of genetic factor gene polymorphisms for NAFLD in older cohorts suggests that age is an important independent risk factor for genetic susceptibility to NAFLD compared to other factors, such as geography, race and ethnicity.
Aging and many aging-related pathological conditions are closely related to mitochondrial function [38].Early senescent cells have also been shown to have high ROS levels, dysfunctional mitochondria and shorter telomeres [39].Moreover, targeted elimination of mitochondria within aging cells has been shown to successfully reverse many features of the aging phenotype, including metabolic disorders [40].Based on the above reports, our study of SAMM50 gene polymorphism in the elderly cohort may better reflect the impact of this mitochondria-related protein on NAFLD.In addition, several studies have found that mitochondrial dysfunction can disrupt hepatic lipid homeostasis, which contributes to the pathogenesis of NAFLD [41][42][43].Specifically, loss of mitochondrial function and subsequent reduction in membrane potential have been reported to disrupt cholesterol homeostasis in macrophages and decrease the efflux of cholesterol to apoA-I [44], while enhancing mitochondrial respiration, as well as ATP production, can increase the expression of the ABCA1 protein and effectively promote cholesterol efflux.The above studies indicate that mitochondria play an important regulatory role in cholesterol metabolism.However, our study found, for the first time, that rs3761472 G carriers had significantly lower TC and LDL than noncarriers.This finding provides compelling evidence for the existence of a potential molecular regulatory mechanism linking age, SAMM50 polymorphisms, mitochondria and cholesterol metabolism.
The analysis of the interaction of multiple SNPs can provide more accurate disease risk prediction models and a more comprehensive understanding of genetic factors' impact on disease, improving disease screening and treatment outcomes in clinical practice.Using the GMDR method, our analysis identified that the three-locus model, comprising rs2073082, rs738491 and rs3761472, provided the most accurate prediction for NAFLD.Although the OR for the risk of NAFLD in rs2073082 G carriers was higher than the OR derived from the three-SNP model, in the screening of the best single locus, rs738491 (TA 0.5260, p 0.3770, CVC 6/10) was superior to rs2073082 and became the best single locus model.The three-locus model, including rs2073082, rs738491 and rs3761472 (TA 0.5552, p 0.0107, CVC 10/10) outperformed the rs738491 single-locus model.In addition, in our cohort, there were 938 carriers of the rs2073082 G allele and 541 with NAFLD.A total of 607 patients had simultaneous mutations at all three loci, of which 351 had NAFLD.Therefore, the inconsistent number of carriers may explain the higher OR of rs2073082 compared to the three genetic loci.However, further confirmation with larger sample sizes is necessary.
GMDR gives the best prediction model based on the accuracy of training and testing, consistency of symbolic testing and CVC.Therefore, it has a relatively reliable reference value.Additionally, rs2073082 G allele, rs738491 T allele and rs3761472 G carriers have a two-fold higher risk of NAFLD compared to noncarriers.Hence, even though rs3761472 showed no connection when examined separately, we assumed that the interplay of the three variations controlled the development of NAFLD.Chen et al. reported a possible synergistic association between rs738491, rs2143571 and rs3761472 of the SAMM50 gene and NAFLD in a middle-aged population [13].The TA of the three-locus model was 60.79% and CVC was 10/10.However, the best model in their study was a two-locus model combining rs2143571 and rs3761472, with the highest TA (62.21%) and CVC (10/10).We conducted a similar study in an elderly population but included a new SNP-rs2073082.Our study found that the three-locus model involving rs2073082, rs738491 and rs3761472 showed the highest TA (55.52%) and perfect CVC (10/10).Differences in the age of the cohort and inclusion of SNPs may account for the differences in the results of the studies.The genetic marker model discovered using GMDR may help clinical identification of potential NAFLD patients, which is expected to become a useful tool for clinical management and lay a foundation for further mechanistic research.
Our study's strength was that it was the first to use a geriatric NAFLD cohort for genetic analysis.Moreover, our study focused on investigating the interaction between three SNPs (rs2073082, rs738491 and rs3761472) and NAFLD in an elderly population.Furthermore, LSM assists in evaluating the relationship between SNPs and liver fibrosis.Nevertheless, there were still some limitations.First, although liver biopsy is considered the gold standard for NAFLD diagnosis, our cohort was built based on a population with a healthy physical examination, and we can only choose B-ultrasound as the first-line diagnostic method.Second, all participants were recruited from the same city in China, and further research work is required to generalize our results to different ethnic groups and the general population in the future.

Conclusions
Together, the three SNPs in the SAMM50 gene provided the most accurate prediction of the predisposition for elderly NAFLD.Among them, the rs2073082 and rs738491 genetic variants contributed to NAFLD susceptibility, whereas the rs738491 T allele and the rs3761472 G allele were linked to fibrosis.Our research uncovered a novel genetic risk factor for elderly NAFLD, which may help to better understand the mechanism.

Supplementary Materials:
The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/biomedicines11092416/s1,Table S1: Multiple comparison correction of LSM based on Table 4; Table S2: Association of SAMM50 rs3761472 G allele carriers and non-carriers with TC, LDL after logistic regression model adjusted for sex, age, BMI and lipid-lowering agent; Table S3: Comparison of clinical characteristics according to SAMM50 genotypes within the whole population; Table S4: Multiple comparison correction of LSM based on Table S3; Table S5: Comparison of clinical characteristics according to SAMM50 genotypes within NAFLD cohorts; Table S6: Comparison of clinical characteristics according to SAMM50 genotypes within NAFLD cohorts (Carriers Vs Noncarriers).

Figure 1 .
Figure 1.Flow chart of patients' selection.A total of 1423 patients were initially collected cohort, 143 were excluded according to the exclusion criteria, and after exclusion of 227 p with incomplete data or unqualified genotyping, 1053 patients were finally enrolled in the st

Figure 1 .
Figure 1.Flow chart of patients' selection.A total of 1423 patients were initially collected in the cohort, 143 were excluded according to the exclusion criteria, and after exclusion of 227 patients with incomplete data or unqualified genotyping, 1053 patients were finally enrolled in the study.

Figure 2 .
Figure 2. SNP−SNP interactions among SAMM50 rs2073082, rs738491 and rs3761472 loci in NAFLD and Control subjects.Inside each square, the left bar represents NAFLD subjects (positive score), and the right bar represents control subjects (negative score).The number at the top of each bar is the scoring statistic, which is the product of the affiliation coefficient and the residuals.The scoring statistic categorizes data into high and low risk by calculating whether an individual's scoring mean exceeds a set threshold (e.g., ≥1).High-risk combinations of genotypes are indicated in dark squares; low-risk genotype combinations are indicated in gray squares; empty squares present the absence of identified genotype combination.

Figure 3 .
Figure 3. SNP−SNP Interaction Dendrogram.Different types of SNP-SNP effects on NAFLD risk.Orange (synergy); Blue (redundancy or antagonism).Short lines represent stronger interactions, and long lines represent weaker interactions.

Figure 2 .
Figure 2. SNP−SNP interactions among SAMM50 rs2073082, rs738491 and rs3761472 loci in NAFLD and Control subjects.Inside each square, the left bar represents NAFLD subjects (positive score), and the right bar represents control subjects (negative score).The number at the top of each bar is the scoring statistic, which is the product of the affiliation coefficient and the residuals.The scoring statistic categorizes data into high and low risk by calculating whether an individual's scoring mean exceeds a set threshold (e.g., ≥1).High-risk combinations of genotypes are indicated in dark squares; low-risk genotype combinations are indicated in gray squares; empty squares present the absence of identified genotype combination.

Figure 2 .
Figure 2. SNP−SNP interactions among SAMM50 rs2073082, rs738491 and rs3761472 loci in NAFLD and Control subjects.Inside each square, the left bar represents NAFLD subjects (positive score), and the right bar represents control subjects (negative score).The number at the top of each bar is the scoring statistic, which is the product of the affiliation coefficient and the residuals.The scoring statistic categorizes data into high and low risk by calculating whether an individual's scoring mean exceeds a set threshold (e.g., ≥1).High-risk combinations of genotypes are indicated in dark squares; low-risk genotype combinations are indicated in gray squares; empty squares present the absence of identified genotype combination.

Figure 3 .
Figure 3. SNP−SNP Interaction Dendrogram.Different types of SNP-SNP effects on NAFLD risk.Orange (synergy); Blue (redundancy or antagonism).Short lines represent stronger interactions, and long lines represent weaker interactions.

Figure 3 .
Figure 3. SNP−SNP Interaction Dendrogram.Different types of SNP-SNP effects on NAFLD risk.Orange (synergy); Blue (redundancy or antagonism).Short lines represent stronger interactions, and long lines represent weaker interactions.

Figure 3 .
Figure 3. SNP−SNP Interaction Dendrogram.Different types of SNP-SNP effects on NAFLD risk.Orange (synergy); Blue (redundancy or antagonism).Short lines represent stronger interactions, and long lines represent weaker interactions.

Figure 4 .
Figure 4. Fruchterman−Rheingold.This interaction model describes the percent of the entropy that is explained by each factor.Each SNP is shown in a box with the percent of entropy below the label.Interactions between SNPs are depicted as lines in different colors.Synergy is depicted as an orange line between SNPs accompanied by a positive percent of entropy, while redundancy is indicated as

Figure 4 .
Figure 4. Fruchterman−Rheingold.This interaction model describes the percent of the entropy that is explained by each factor.Each SNP is shown in a box with the percent of entropy below the label.Interactions between SNPs are depicted as lines in different colors.Synergy is depicted as an orange line between SNPs accompanied by a positive percent of entropy, while redundancy is indicated as a blue line accompanied by a negative percent of entropy.Rs3761472 synergized with both rs738491 and rs2073082, resulting in positive information gain values of 0.12% and 0.04% in NAFLD, while rs2073082 and rs738491 were antagonistic with negative IG values (−0.76%).
X. and S.Z.; Supervision, Y.Z. and J.Z. (Jing Zhang); Writing-original draft, J.Z. (Jinhan Zhao); Writing-review & editing, J.Z. (Jinhan Zhao), Y.Z. and J.Z. (Jing Zhang).All authors have read and agreed to the published version of the manuscript.Funding: This work was supported by the National Natural Science Foundation of China (82070627); the Beijing Municipal Natural Science Foundation (7222090); the Beijing Municipal Institute of Public Medical Research Development and Reform Pilot Project (2021-10); the Scientific Research Project of Beijing Youan Hospital, CCMU, 2022 (BJYAYY-YN2022-17); and the Jiangxi National Science Foundation (No. 2020BABL206092).Capitals's Funds for Health Improvement and Research (2022-2Z-2187).Institutional Review Board Statement: The study protocol was approved by the Ethics Committee of Beijing You'an Hospital, Capital Medical University (IRB number [2020]-133).The approval date was 28 October 2020.The registration number was ChiCTR 2100043106.All study subjects provided written informed consent.

Table 1 .
Clinical characteristics of non-NAFLD controls and NAFLD patients.

Table 2 .
Distribution of genotypes and allele frequencies of three genetic variants in SAMM50 in subjects.

Table 3 .
Study group odds ratios for NAFLD according to genotypes of SAMM50 single nucleotide polymorphisms in the study group.
a Multiple logistic regression model was adjusted for age, gender, and body mass index.

Table 4 .
Comparison of clinical characteristics according to SAMM50 genotypes within the whole population (Carriers vs. Noncarriers).

Table 5 .
Best models to predict NAFLD by generalized multifactor dimensionality reduction (GMDR) a .

Table 6 .
Logistic regression analysis between each genetic model and risk of NAFLD.
a Multivariate logistic regression model adjusted for age, sex and BMI.BMI, body mass index; CI, confidence interval; OR, odds ratio.