Apolipoprotein E Gene Variants and Risk of Coronary Heart Disease: A Meta-Analysis

Objectives. Apo E genes involved in lipoprotein synthesis and metabolism are considered one of the candidates to CHD. However, the results remain conflicting. Methods. We performed this meta-analysis based on 30 published studies including 11,804 CHD patients and 17,713 controls. Results. Compared with the wild genotype E3/3, the variant genotypes ApoEE3/4 and E4/4 were associated with 22% and 45% increased risk of CHD, respectively (E3/4 versus E3/3: OR = 1.22, 95% CI = 1.15–1.29; E4/4 versus E3/3: OR = 1.45, 95% CI = 1.23–1.71). Besides, compared with ε3 allele, carriers with the ε4 allele had a 46% increased risk of CHD (OR = 1.46, 95% CI = 1.28–1.66), while the ε2 had no significantly decreased risk of CHD. In the subgroup analysis by ethnicity, ε4 had a 25% increased risk of CHD in Caucasians (OR = 1.25, 95% CI = 1.11–1.41), and the effects were more evident in Mongolians (OR = 2.29, 95% CI = 1.89–2.77). The ε2 allele had a decreased risk of CHD in Caucasians (OR = 0.84, 95% CI = 0.74–0.96), but not in Mongolians. Conclusions. The analysis suggested that ApoEε4 mutation was associated with the increased risk of CHD, while ApoEε2 allele had a decreased risk of CHD just in Caucasians.


Introduction
Coronary heart disease (CHD) as a multifactorial disease caused by genetic and environmental factors became one of the leading causes of mortality and morbidity worldwide, especially in the developed countries [1]. Previous studies provided evidence that risk factors for CHD, including diabetes mellitus, smoking, and arterial hypertension, would contribute to rapid process of clinical events such as myocardial infarction (MI), ischemic heart failure, and death [2,3]. Moreover, apart from the above risk factors, population-based studies have reported that genetic susceptibility accounts for around 50% of the risk for CHD, which suggested that the host genetic background plays an important role in the onset and development of CHD as well [4,5]. The determination of blood lipid and lipoprotein levels is one of the coronary risk factors. Apolipoprotein genes involved in lipoprotein synthesis and metabolism play imperative roles in studying the susceptibility to CHD and cerebrovascular disease [6][7][8][9][10][11].
In 1992, Dallongeville et al. tested the consistent relationship between plasma TG levels and Apo E phenotype among 45 different populations in a meta-analysis [12]. Another meta-analysis including 14 studies showed that subjects with 4 allele were associated with 26% increased risk of CHD compared with 3 allele [13]. Since the publication of these two meta-analyses, numerous studies have appeared in recent years [14][15][16][17]. However, differences in study design, end point validation, choice of subjects, and limited statistical power have led to different results of Apo E genotypes on CHD risk in the general population. Therefore, the present metaanalysis is designed to derive a more plausible estimation.

Quality Score Assessment.
To determine the methodological quality of each study, we used the Newcastle-Ottawa scale (NOS). The NOS ranges between zero (worst) up to nine stars (best). Two authors of this article independently assessed the qualities of included studies. Disagreement was resolved by discussion.

Meta-Analysis.
The risks (odds ratios, ORs) of CHD associated with the Apo E polymorphisms were estimated for each study with the software Stata12.0. The risk of the variant genotypes E2/2, E2/3, E2/4, E4/3, and E4/4 was estimated compared with the genotype E3/3 homozygotes. In addition to comparisons for total subjects, studies were categorized into different subgroup analyses according to the ethnicity. We estimated the between-study heterogeneity across the eligible comparisons using the Cochrane Q-test and the heterogeneity was considered significant for < 0.1. Fixed effect or random effect was used to calculate pooled effect estimates. Random effects incorporate an estimate of the between-study variance and tend to provide wider confidence intervals, when the results of the constituent studies differ among themselves. In the absence of between-study heterogeneity, the two methods provide identical results.
Publication bias was evaluated by funnel plot and Begg's and Egger's tests. The Hardy-Weinberg equilibrium (HWE) was tested by a goodness-of-fit 2 test to compare the observed genotypes frequencies with the expected ones Records after duplicates removed (n = 576) among control subjects. For sensitivity analyses, we examined whether the excluding studies with substantial deviation from HWE among controls affected our pooled estimates of ORs. Finally, we use the following formula to estimate the failsafe number: is the number of included studies and is the value of the independent study. The result is obtained from the software SAS 9.2.

Studies
Characteristics. 714 studies were searched, among which 30 studies were included in the final meta-analysis . The study selection process is detailed in Figure 1.
Given in Table 1 were the lists of number of cases, controls, HWE, and the NOS score of these 30 case-control studies. All studies indicated that the distribution of genotypes in the controls was consistent with HWE, except for 7 studies [29,35,37,39,40,45,46] ( < 0.05). According to the quality criteria, there were 21 studies with high quality (NOS score > 6). Tables 2 and 3 showed the frequency distributions of the Apo E alleles and genotypes in the cases and controls. Figures 2 and 3 showed the ORs on CHD associated with ApoE 2 alleles and ApoE 4 alleles compared with the 3 alleles in individual studies.  Besides, carriers with ApoE 2 allele had no significantly decreased risk of CHD compared with individuals with the 3 allele in the random-effect model (OR = 0.91; 95% CI = 0.81-1.03). Stratified analysis on the descent also showed no evidence on the 2 allele variant and CHD risk in Mongolians (OR = 1.18, 95% CI = 0.94-1.46), but there had a decrease risk in Caucasians (OR = 0.84, 95% Cl = 0.74-0.96). In addition, compared with the ApoE 3 allele, carriers with the 4 allele had a 46% increased risk of CHD in the random-effect model

Sensitivity Analysis.
We conducted a sensitivity analysis on the Apo E polymorphisms and risk of CHD excluding studies deviating from HWE among controls. The pooled ORs estimates were similar with that of excluded studies, so the results were not shown.

Bias Diagnostics.
For the ApoE 2 versus 3 allele, the shape of the funnel plot seemed symmetrical (Figure 4), and then Egger's test showed no evidence of publication bias ( = 0.211), and the fail-safe number also showed no publication bias in this meta-analysis (N fs (0.05) = 381.51, N fs (0.01) = 173.87).
For the ApoE 4 compared with 3 allele, the shape of the funnel plots seemed asymmetrical (Figure 4), and Egger's test revealed there was a significant publication bias ( = 0.003). By using the trim and fill method, we showed that OR and 95% CI did not change. Besides, the fail-safe number also suggested that the result of the meta-analysis was stability (N fs (0.05) = 1218.31, N fs (0.01) = 588.44).

Discussion
Apolipoprotein E (Apo E) is one of the most major apolipoproteins in the central nervous system, with functions of neurons repair. Apo E genetic variants showed significant    associations with the risks of nervous system degenerative diseases, including Alzheimer's disease, vascular dementia, and cerebrovascular disease [48,49]. Apo E gene as a receptor-binding ligand mediating the clearance of chylomicron and remnants of very-low-density lipoprotein cholesterol from plasma also plays a major role in the metabolisms of cholesterol and triglyceride. Functional variants of genes encoding lipoproteins are responsible in part for betweenindividual variation in the plasma levels of lipoproteins.
A lot of epidemiologic studies have investigated the relation between Apo E genotypes and CHD risk in the general population. Apo E polymorphisms are believed to confer susceptibility to CHD risk. In 1992, a meta-analysis of 27 studies reported that the subjects carrying the 2 and 4 alleles had, respectively, lower and higher plasma cholesterol values than subjects carrying the E3/3 genotype, which suggested that the ApoE 4 allele may, in individuals with the ApoEE4/3 phenotype, be a risk factor of cardiovascular disease [12]. The last meta-analysis of 14 published case-control studies in 2015 showed that carriers with POE 2 allele were associated with the decreased risk for CHD (OR = 0.82, 95% CI: 0.75-0.90) compared with those carrying 3 allele, while those with 4 allele had a significant increased risk for CHD (OR = 1.34, 95% CI: 1.15-1.57) [15].
In this 30 studies' meta-analysis including 11,804 CHD patients and 17,713 controls, we identified a significant increased risk for CHD among carriers of the ApoEE3/4 and E4/4 genotypes compared with carriers of the E3/3 genotype, but no significant evidence was found between the variant genotypes of ApoEE2/3, E2/4, and E2/2 and CHD risk. In the stratified analyses by descent, for the 4 allele genetic variant, the effect was more evident in the Mongolians group and mild in Caucasians group, which showed that the ApoE 4 allele genetic variant modulated the increased risk of CHD with the differences of genetic background. Moreover, there was a decreased risk in Caucasians between the 2 allele variant and CHD risk but no evidence in Mongolians; further studies should be conducted to verify it.
Our study has several limitations. First, as with all metaanalyses, although we did Egger's test and calculated the failsafe number to evaluate the publication bias, it might have occurred because our analyses were all based on published studies. For the ApoE 4 compared with 3 allele, the Egger test showed existence of publication bias, but from the results of the trim and fill method and the fail-safe number, the publication bias and the possibility of false positive were relatively small. Second, the control group of some studies was not in conformity with Hardy-Weinberg equilibrium. But, in sensitivity analysis, the pooled estimates were similar after we excluded studies deviating from Hardy-Weinberg equilibrium among controls; therefore these studies were included in the final analysis. Gene-environment interactions may have contributed to the CHD. Apo E gene is a candidate gene and a common one to study gene-environment interactions. However, because of lack of the original data of the meta-analysis, further evaluation of potential gene-gene and gene-environment interactions was limited.
In conclusion, it was showed in this meta-analysis that ApoE 4 allele polymorphism may contribute to the individual susceptibility of CHD. Further rigorous design, large sample of case-control, or prospective study are required to continue in-depth evaluation on gene-gene and gene-environment interactions on Apo E polymorphisms and CHD risk.