A systematic review and meta-analysis of observational studies on the effects of epigenetic factors on serum triglycerides

ABSTRACT Epigenetic modifications might be associated with serum triglycerides (TG) levels. This study aims to systematically review the studies on the relationship between the methylation of specific cytosine-phosphate-guanine (CpG) sites and serum TG levels. This systematic review and meta-analysis study was conducted according to the PRISMA 2020 (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement. A systematic literature search was conducted in Medline database (PubMed), Scopus, and Cochrane library up to end of 2020. All observational studies (cross-sectional, case-control, and cohort) were included. Studies that assessed the effect of DNA methylation of different CpG sites of ABCG1, CPT1A, and SREBF1 genes on serum TG levels were selected. The National Institutes of Health (NIH) checklist was used to assess the quality of included articles. Among 2790 articles, ten studies were included in the quantitative analysis and fourteen studies were included in the systematic review. DNA methylation of ABCG1 gene had significant positive association with TG levels (β = 0.05, 95% CI = 0.04, 0.05, P heterogeneity < 0.001). There was significant inverse association between DNA methylation of CPT1A gene and serum TG levels (β = −0.03, 95% CI = −0.03, −0.02, P heterogeneity < 0.001). DNA methylation of SREBF1 gene was positively and significantly associated with serum TG levels (β = 0.03; 95% CI = 0.02-0.04, P heterogeneity < 0.001). DNA methylation of ABCG1 and SREBF1 genes has positive association with serum TG level, whereas this association is opposite for CPT1A gene. The role of epigenetic factors should be considered in some populations with high prevalence of hypertriglyceridemia.


INTRODUCTION
E levated level of serum triglycerides (TG) is one of the major components of the metabolic syndrome. It is associated with adverse health effects including cardiovascular disease (CVD), obesity, and insulin resistance (1)(2)(3). Some evidence has recommended that hypertriglyceridemia should be managed in any severity (1,4). The prevalence of hypertriglyceridemia has large variations in different population and is highly influenced by different variables such as gender and age (5,6).
Approximately one fifth of adult females and about one third of adult males have hypertriglyceridemia in Spain (6). A cross-sectional study reported that the mean TG level has decreased from 2007 to 2014 in the US (7). Based on cross-sectional studies in Indian population over 20-year period, prevalence of hypertriglyceridemia has increased from 25.7% to 32.8% (8). Some factors including lifestyle, environmental and genetic factors influence on serum lipid profiles (9). The role of epigenetic factors needs to be clarified in this regard.
Epigenetic processes are natural and require for functions of various organisms. However, their improper occurrence will have detrimental impacts on health and behavioral. Epigenetic changes are DNA changes without any effect on DNA sequences but impact on gene expression and activity (9,10). Several types of epigenetic processes have been identified that DNA methylation is one of the most important epigenetic changes. The covalent transfer of a methyl group to cytosine-guanine (CpG) dinucleotides of DNA by DNA methyltransferase leads to DNA methylation (11,12). CpG dinucleotides are often located in regulatory parts of genes, so they can regulate genes expression (13).
Several studies have assessed the relationship between DNA methylation at different loci of different genes with serum TG levels, and had inconsistent findings. Some results did not show significant association between DNA methylation and TG levels (11,13), whereas some of them confirmed this association (1,9,(14)(15)(16). Different sample size, gene, method, and design might explain inconsistent findings. There are several epigenetic changes that might affect TG levels; some genes including ABCG1, CPT1A and SREBF1 have been assessed more than others. The contradictory findings of various studies show the importance of providing a holistic overview. Therefore, this systematic review and meta-analysis aims to provide a summary of the literature that have evaluated the relationship between methylation of different CpGs of ABCG1, CPT1A, SREBF1 genes and serum TG levels.

Search strategy
The current systematic review and meta-analysis study was conducted according to the PRISMA 2020 (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement (17). The protocol was registered on PROSPERO (ID: CRD42020202332). A systematic literature search was conducted in Medline database (PubMed), Scopus, and Cochrane library up to end of July 2020. Databases were searched daily for any newly published articles up to end of December 2020, and updated till September 2021. The following search terms were used: (Triglyceride OR TG OR Triacylglycerol OR Hypertriglyceridemia) AND (Epigenetic OR Epigenomic OR methylation OR "DNA methylation" OR acetylation OR "DNA acetylation"). The search string for each database was summarized in the supplementary file 1.
Two reviewers (SMT and MHB) independently reviewed and screened the appropriate published papers based on title, abstract, and full text. In addition, the reference lists of related review articles were checked to find undetected relevant studies. Any discrepancy related to eligible records was resolved by the third reviewer (RK).

Inclusion criteria
Only English-language articles and human studies were included. We included studies that extracted DNA from blood samples and measured DNA methylation. All observational studies (cross-sectional, case-control, and cohort) on individuals over 18 years of age that assessed the effect of epigenetic changes on serum TG levels were included without restriction of gender, race, ethnicity, and year of publication.

Exclusion criteria
Those papers with the following criteria were excluded: duplicate publications and studies that assessed the effect of hypertriglyceridemia on epigenetic changes.

Data extraction
The number of studies on some genes including ABCG1, CPT1A, and SREBF1 were greater than others. So, we selected them for the meta-analysis. Data extraction was conducted by two reviewers (SMT and MHB) independently and was checked by the third reviewer (RK). The following information was extracted from eligible studies: first author, publication year, study design, sample size, age, body mass index (BMI), the gene(s) and loci and type of tissue sample.

Quality assessment
In order to evaluate the risk of bias of 14 included studies in the systematic review, two reviewers (SMT, MHB) independently assessed the quality of the articles by National Institutes of Health (NIH) Quality Assessment Tool (18). All included studies had cohort and cross-sectional design. Any disagreement was resolved by consulting with the third researcher (RK). The scale consist of 14 questions. Each item was answered as "yes", "no", "not applicable" or "not reported". Table 1 shows the quality assessment of included articles.

Statistical analysis
The β-coefficient values of selected studies were applied for pooled analysis. The potential heterogeneity across studies was evaluated using the Cochran's Q-test and was expressed using the I 2 index. The pooled results were calculated by the random-effects model. Subgroup analyses based on CpG sites were performed to seek the sources of heterogeneity. In addition, meta-regression was used for assessing the mean age, mean BMI, sample size and the year of publication of studies as the possible source of heterogeneity. The sensitivity analyses were performed by excluding one study at a time to gauge the robustness of our results. Publication bias was evaluated by Funnel plot and Egger's test. The possible publication bias was adjusted using the trim and fill method. Statistical analyses were conducted using the STATA 12.0 software (STATA Corp, College Station, Texas, USA). P < 0.05 was considered as significance level.

Study selection
The flow diagram for the process of study selection is shown in Figure 1. The initial search recognized 2,790 articles and 2510 of them remained after excluding duplicates. After screening the title and abstracts, 2,426 articles were excluded, and 84 articles remained for further assessment. The full texts of remaining studies were reviewed carefully by two researchers. Any discrepancy was resolved by the third reviewer. Finally, 14 articles were included in the systematic review, and 10 of them were included in the meta-analysis.

Study characteristics
Fourteen eligible studies were included in the systematic review. Table 2 shows the general characteristics of the included studies.
Association between DNA methylation of the ABCG1 gene and serum TG levels The findings of meta-analysis on 8 studies showed that DNA methylation of ABCG1 gene had significant positive association with serum TG levels (β = 0.05; 95% CI [0.04, 0.05]) using random effect model. There was significant heterogeneity (P < 0.001), with I 2 values of 95%. Therefore, the subgroup analysis, metaregression and sensitivity analysis were used to explore the potential sources of heterogeneity ( Figure 2).

Subgroup analysis
Results of the following sites, cg07397296, cg01881899, cg02370100, ABCG1-CPG3. So we considered these sites as one group. The heterogeneity was significant for each three groups of CpGs ( Figure 2).

Sensitivity analysis
Results of sensitivity analysis showed that the pooled effect size (β) and heterogeneity did not influence after excluding studies one by one. After excluding the study of Guay and cols. (

Publication bias
Funnel plot showed asymmetry. The P-value for Egger's test was < 0.0001, which revealed obvious publication bias among studies. Therefore, trim and fill analysis was performed and the pooled effect size was obtained (β = 0.033 (95% CI: 0.026, 0.039); number of trimmed studies: 10).
Association between DNA methylation of the CPT1A gene and serum TG levels Results of meta-analysis on 7 studies indicated that the DNA methylation of CPT1A gene had significant inverse association with serum TG levels (β = -0.03 [95% CI: -0.03, -0.02]) using the random effects model. The heterogeneity was significant (I 2 = 93.5%; P < 0.001). . Two studies had the relevant effect size for cg09737197, and one study had the relevant effect size for cg01082498, so we considered these sites as one group. The heterogeneity was significant for each three groups of CpGs (p < 0.001). The results of metaregression analysis showed that mean age, mean BMI, sample size and the year of publication had no significant effect on the association between DNA methylation of CPT1A gene and serum TG levels (p > 0.05) (Figure 3). Results of sensitivity analysis showed that the pooled effect size (β) was not influenced after excluding studies one by one. Funnel plot was asymmetry. The P-value for Egger's test was < 0.001, thus there was obvious publication bias among these studies. Trim and fill analysis were applied, but no studies were filled. It showed that the publication bias had a non-significant effect on the results.
Association between DNA methylation of the SREBF1 gene and serum TG levels and also for other group (cg20544516, cg08129017) (β = 0.02; 95% CI [0.01, 0.03, I 2 = 84.6%]) were positive and significant (Figure 4). One study had the relevant effect size for cg20544516, and one study had the relevant effect size for cg08129017. So, we considered these sites as one group. The heterogeneity was significant for both groups. The results of metaregression analysis indicated that mean age, mean BMI, sample size and year of publication had no significant association with the effect of SREBF1 gene on serum TG levels (p > 0.05). Results of sensitivity analysis showed that the pooled effect size (β) was not influenced after dropping studies one by one. Funnel plot was asymmetry. The P-value for Egger's test was 0.001. Therefore, there was publication bias among these studies. So, trim and fill analysis was performed and the pooled effect size obtained (β = 0.015 [95% CI: 0.007, 0.023]); number of trimmed studies: 6).

DISCUSSION
The present meta-analysis investigated the association between DNA methylation of different CpG sites of ABCG1, CPT1A, SREBF1 and serum TG levels. Higher methylation of ABCG1 and SREBF1 genes, and lower methylation of CPT1A gene were significantly associated with higher serum TG levels. These findings can explain the reason of the large variations in the prevalence of hypertriglyceridemia in different populations.
The current meta-analysis showed that methylation in cg06500161 and cg27243685 sites has greater impact on serum TG levels. Most of the included studies revealed significant positive association between ABCG1 methylation and serum TG levels (1,9,12,14,15,19,23). While, one study on 139 individuals with coronary heart disease in China, did not show any significant association between ABCG1 methylation and serum TG level (13). Another study on patients with familial hypercholesterolemia found significant association between ABCG1 DNA methylation and HDL-C, LDL-C and TG levels only in women (11).
Study on 3296 participants from six cohorts in the Netherland suggested that the levels of blood TG influenced on the ABCG1 methylation. They did not observe any causal effects of DNA methylation on lipid profiles (26). Studies demonstrated an association between higher ABCG1 methylation and an increased risk of coronary heart disease, type 2 diabetes, and metabolic syndrome (13,15,19,27). Two populationbased cohorts on 2306 individuals indicated that higher methylation at cg27243685 site of ABCG1 gene increased the serum TG level and was associated with an increased risk of coronary heart disease (14).
Dekkers and cols. suggested that blood TG level had causal effect on the level of CPT1A methylation but not vice versa (26). Another study on 992 participants in USA showed that both CPT1A methylation and blood TG levels might have reciprocal causal effect. However, because of the small sample size and lack of strong genetic instruments, they could not determine the direction Epigenetic and serum triglycerides Arch Endocrinol Metab. 2022;66/3 association between CPT1A methylation and serum TG levels (25). It is suggested that CPT1A expression might act as a useful biomarker for CVD (28).

SREBF1
(sterol regulatory element-binding transcription factor 1) gene is located on chromosome 17 and encodes sterol regulatory element binding proteins (SREBPs) (22). SREBP-1a and SREBP-1c are two isoforms of SREB proteins that are encoded by SRBF1 gene. SREBF-1a modulates the expression of cholesterol and fatty acid biosynthesis genes, andSREBF-1c regulates the expression of fatty acid, phospholipid, and TG biosynthetic genes. Studies demonstrated that variation in lipid profiles could be related to these proteins (32,33). According to the current literature, the level of methylation at three loci of SREBF1 gene including: cg11024682 (9,12,14,15,(23)(24)(25), cg20544516 (15) and cg08129017 (14) has significant and positive association with serum TG level. There is an inverse association between SREBF1 methylation and its expression. Thus, decreasing the expression of SREBF1 might be the reason of hypertriglyceridemia (9,23). The current meta-analysis shows the effect of cg11024682 on serum TG levels is greater than other sites of SREBF1. Likewise, a study on 1,408 men and 1888 women of six cohorts showed that higher serum TG levels was correlated with higher methylation of cg11024682 (26). MicroRNA33b (MIR33b) is encoded by cg20544516 site on SREBF1 gene. MIR33b suppress several genes such as ABCG1 and CPT1A that they contribute to oxidation and transport of fatty acid (15,34,35).
Recent studies demonstrated that the methylation of ABCG1, CPT1A, and SREBF1 genes was associated with some cardio-metabolic risk factors including total cholesterol and insulin levels, as well as anthropometric indices of general and abdominal obesity. Therefore, study on these genes can be important in prevention of non-communicable diseases (12,15,23).
According to the current findings, DNA methylation affects gene expression and new therapeutic pathways could be appeared for management of hypertriglyceridemia. Some medications including fenofibrate might affect serum TG levels through altering DNA methylation (36). However, a study with short-time follow up could not confirm these findings (37).
The present study has some limitations. First, the selection bias was unavoidable. Second, heterogeneity existed in our meta-analysis, and it could not be solved by applying any statistical method, differences in the genetic methods, age, BMI, or race of participants in various studies might be the reasons of this heterogeneity. The main strength of our study is its novelty; to the best of our knowledge, no previous systematic reviews and meta-analysis have specifically evaluated the relationship between methylation of different CpGs of some genes and serum TG levels. Meta-regression analysis was used to find potential variables as the main source of the heterogeneity.
In conclusion, the results of the present metaanalysis show that methylation of ABCG1 and SREBF1 genes have positive association with serum TG levels, whereas this association is inverse for methylation of CPT1A gene. DNA methylation on cg06500161 and cg27243685 at ABCG1 gene, cg00574958 and cg17058475 at CPT1A gene and cg11024682 at SREBF1 gene are correlated with serum TG levels more than other CpGs. Although there are several studies that assessed the effect of DNA methylation on serum TG levels, but limited experience exists on the impact of serum TG levels on DNA methylation. Further studies with long follow up periods are needed to assess the causal effect of DNA methylation on serum TG levels and vice versa.
The role of epigenetic factors should be considered as one of underlying factors for the considerable variations in the prevalence of hypertriglyceridemia among different populations. Author contribution: SMT, MHB and RK contributed to the conception and design of the research. SMT, MHB,PP and RK reviewed the literature, and drafted the manuscript. MK contributed to the analysis of data. SMT, MHB, RK, and PP contributed to the interpretation of the data and revision. All authors read and approved the final manuscript, and accept its content.
Disclosure: no potential conflict of interest relevant to this article was reported. Figure S1. Meta-regression plot of the impact of age, BMI, year and sample size of studies on the association between ABCG1 gene and TG levels, the size of the circles is depended on the weight of the study in the fitted random-effects meta-regression model.  Figure S2. Meta-regression plot of the impact of age, BMI, year and sample size of studies on the association between CPT1A gene and TG levels, the size of the circles is depended on the weight of the study in the fitted random-effects meta-regression model. Figure S3. Meta-regression plot of the impact of age, BMI, year and sample size of studies on the association between SREBF1 gene and TG levels, the size of the circles is depended on the weight of the study in the fitted random-effects meta-regression model.