Association between apolipoprotein gene polymorphisms and hyperlipidemia: a meta-analysis

Background Hyperlipidemia plays an important role in the etiology of cardio-cerebrovascular disease. Over recent years, a number of studies have explored the impact of apolipoprotein genetic polymorphisms in hyperlipidemia, but considerable differences and uncertainty have been found in their association with different populations from different regions. Results A total of 59 articles were included, containing in total 13,843 hyperlipidemia patients in the case group and 15,398 healthy controls in the control group. Meta-analysis of the data indicated that APOA5–1131 T > C, APOA1 -75 bp, APOB XbaI, and APOE gene polymorphisms were significantly associated with hyperlipidemia, with OR values of 1.996, 1.228, 1.444, and 1.710, respectively. All P-values were less than 0.05. Conclusions Meta-analysis of the data indicated that the C allele of APOA5 1131 T > C, the A allele at APOA1-75 bp, the APOB XbaI T allele, and the ε2 and ε4 allele of APOE were each a risk factor for susceptibility for hyperlipidemia.


Background
Cardio-cerebrovascular disease is the leading cause of increased human mortality, globally [1]. Recently, studies have shown that the fatality rate from cardiocerebrovascular disease accounts for approximately 30% of the total global death toll [2]. Hyperlipidemia is a chronic non-communicable disease caused by an imbalance in the structure of plasma lipids caused by a fat metabolism disorder [3]. It is the primary risk factor for atherosclerosis and the pathological basis for cardio-cerebrovascular disease [4]. In addition, a large number of manuscripts have demonstrated that hyperlipidemia is a pathogenic factor of digestive and urinary diseases such as diabetes, hepatopathy, and pancreatitis. Hyperlipidemia can be categorized as hypercholesteremia, hypertriglyceridemia, mixed hyperlipidemia, and low-density lipoproteinemia, etc. Medical research has established that the mechanism of hyperlipidemia is not only determined by environmental factors, such as long-term consumption of large quantities of saturated fatty acids, cholesterol, and sugar, it is also influenced by genetic factors at gene loci. There are multiple academic reports that apolipoprotein (APO) gene mutations are closely related to disorders of blood lipid metabolism [5]. APO is an important component of lipoprotein. So far, more than 20 forms of APO have been identified, including APOA, APOB, APOC, APOD, APOE, APOH, APOM, etc. [6] Single nucleotide polymorphisms (SNPs) are changes to a single nucleic acid in a protein caused by the insertion, deletion, or substitution of a single nucleotide base in the gene sequence. Of the existing apolipoprotein candidate genes, researchers have correlated APOA1, APOA5, APOB, and APOE gene polymorphisms with hyperlipidemia. APOA1 and APOA5 genes are located in the long arm region of chromosome 11. APOA1 is located in the APOA1-C3-A4 gene cluster, the principal site controlling the expression of lipids and lipoproteins [7]. APOA5 is located downstream of APOA4, and its distance from the APOA1/C3/A4 gene cluster is approximately 30 kb. The APOA5 gene is most commonly altered at -1131 T > C, this polymorphism being closely associated with a number of diseases, such as hypertriglyceridemia and coronary heart disease [8]. The APOB gene is located in the short arm of chromosome 2 and contains 29 exons and 28 introns. The cleavage sites MspI and XbaI are located within exon 26 of the APOB gene. The EcoRI cleavage site is located within exon 29 [9]. A number of studies have clearly indicated that the APOB gene affects lipid metabolism to a certain extent. The APOE gene is located on chromosome 19 with a polymorphic gene structure. The isomers are encoded by the three alleles ε2, ε3, and ε4 [10], forming 6 genotypes E2/2, E3/3, E4/4, E2/3, E2/4, and E3/4, of which E3/3 is the most common within the population.
Over recent years, there have been multiple studies that have explored the correlation between genetic polymorphism and hyperlipidemia for the apolipoprotein gene loci described above, but there are great differences and uncertainties in different populations from different regions. Therefore, in the present review, we systematically searched the literature and reviewed case-control studies of hyperlipidemia. A meta-analysis was conducted to explore the relationship between APOA (A1-75bp, A1 + 83 bp, A5-1131T>C), APOB (MspI, XbaI, EcorI), and APOE with hyperlipidemia so that an evidence-base can be provided for the prevention and control of hyperlipidemia.

Study characteristics
A total of 3706 articles were identified in the Chinese and English databases, of which 59 articles were finally selected, including 22 that analyzed APOA, 28 APOB, and 30 APOE. Three sites in the APOA gene were studied: A5-1131T > C was studied in 10 case-control studies that included 1211 cases and 1495 controls; A1-75bp was studied in 5 case-control studies that included 1284 cases and 1312 controls; and A1 + 83 bp was studied in 7 case-control studies that included 1452 cases and 1620 controls. The APOB gene was investigated at three sites: MspI was studied in 6 case-control studies that included a hyperlipidemia group, with 1155 cases and 1043 controls; XbaI was studied in 12 case-control studies that included 1900 cases and 1836 controls; and EcorI was studied in 10 case-control studies that included 1633 cases and 1686 controls. The APOE gene is co-coded by the three alleles, ε2, ε3, and ε4, for which 30 case control studies were studied that included 5208 cases in the hyperlipidemia group and 6406 cases in the control group. The NOS score of no study included in the review was less than 7. The comparison between case and control groups was highly credible. The specific process for literature retrieval is displayed in Fig. 1.

Meta-analysis of APOA5-1131 T > C (rs662799)
This gene locus was included in 10 case-control studies, involving a total of 2706 subjects, including 1211 in the hyperlipidemia group and 1496 in the control group. The baseline data and quality evaluation of each study are displayed in Table 1. Analysis of the relationship between C vs T alleles and hyperlipidemia (allele model) revealed substantial heterogeneity (I 2 = 73.9%, P < 0.001), so a random-effects model was used to analyze the combined effects. Individuals with the C allele had a higher risk of hyperlipidemia than those with the T allele, a difference that was statistically significant (OR = 1.996, 95% CI = 1.529-2.606, P < 0.001) (Fig. 2). Other gene models at this site displayed consistent results (Table 2). Subgroup analysis by ethnicity demonstrated an increased risk of hyperlipidemia among Asians (OR = 1.818; 95% CI = 1.268-2.607, P = 0.001) and Caucasians (OR = 2.355; 95% CI = 1.665~3.331, P < 0.001) that had the C allele, using the allele model. Other gene models at this site displayed results that were consistent with this (Table 3, Fig. 3). Therefore, the single nucleotide polymorphism APOA5-1131 T > C was associated with hyperlipidemia, the C allele posing a risk factor for susceptibility to hyperlipidemia.

Meta-analysis of APOA1-75 bp (rs670)
This location on APOA was included in 5 case-control studies, involving a total of 2596 subjects, of which 1284 were in the hyperlipidemia group and 1312 in the control group. Baseline data and quality evaluation are displayed in Table 1. There was no significant heterogeneity in the relationship between A vs G alleles and hyperlipidemia (allele model) (I 2 = 1.2%, P = 0.400), and so a fixed-effects model was used to combine the effects. Individuals with the A allele had a higher risk of hyperlipidemia than those with the G allele, a difference that was statistically significant (OR = 1.228, 95% CI = 1.067-1.413, P = 0.004) (Fig. 4). The recessive model of this locus indicated that the difference was not statistically significant (P = 0.066). Other gene models at this site were consistent with this result, suggesting that the single nucleotide polymorphism APOA1-75 bp is associated with hyperlipidemia, the A allele being a risk factor for susceptibility to hyperlipidemia (Table 2).

Meta-analysis of APOA1 + 83 bp (rs5069)
This site was included in 7 case-control studies, involving a total of 3072 subjects, including 1452 in the hyperlipidemia group and 1620 in the control group. The baseline data and quality evaluation of each study are shown in Table 1. Analysis of the relationship between A vs G alleles and hyperlipidemia (allele model) indicated that there was no significant heterogeneity (I 2 = 0.0%, P = 0.472). Therefore, a fixed-effects model was selected to analyze the pooled effect. There was no significant difference in risk in individuals that carried the T allele compared with C (OR = 0.928, 95% CI = 0.771-1.116, P = 0.425). The P-values of other gene models at this locus were all higher than 0.05, suggesting that there was no significant difference. Thus, an association between APOA1 + 83 bp gene polymorphism and susceptibility to hyperlipidemia can be considered not to exist ( Table 2).

Meta-analysis of APOB MspI (rs1801701)
This gene locus was included in 6 case-control studies, involving a total of 2198 subjects, including 1155 in the hyperlipidemia group and 1043 in the control group. Baseline data and quality evaluation are shown in Table 4. Analysis of the association between M-vs M+ alleles and hyperlipidemia (allele model) indicated no heterogeneity (I 2 = 0.0%, P = 0.731), and do a fixedeffects model was selected to analyze the pooled effects. No significant difference in risk was found in individuals carrying the M-compared with the M+ allele (OR = 0.892, 95% CI = 0.756-1.053, P = 0.178). The P-values of other gene models at this site were also greater than 0.05, indicating that there was no significant difference. Thus, no association between genetic polymorphism of APOB MspI and risk of hyperlipidemia was found ( Table 5).

Meta-analysis of APOB XbaI (rs693)
This site was included in 12 case-control studies, involving a total of 3736 subjects, including 1900 in the hyperlipidemia    Table 4. Analysis of the association between T vs C alleles and hyperlipidemia (allele model) indicated substantial heterogeneity (I 2 = 72.4%,P < 0.001) and so a random-effects model was used to analyze the pooled effects. The risk of hyperlipidemia in the T allele population was higher than that with the C allele, the difference of which was statistically significant (OR = 1.444, 95% CI = 1.061-1.966, P = 0.020) (Fig. 5). There was no significant difference between the dominant and codominant models of this locus, with P-values of 0.100 and 0.140, respectively. The results of other gene models were consistent with those of the allele model (Table 5). Subgroup analysis by ethnicity displayed an increased risk of hyperlipidemia among Caucasians that carried the T allele when analyzed with the allele model, a difference that was statistically significant (OR = 2.074; 95% CI = 1.611-2.672, P < 0.001). However, no significant association was found in other gene models. We found that there was no significant association with risk of hyperlipidemia risk in Asians carrying the T   Fig. 6). Therefore, an association between APOB XbaI gene single nucleotide polymorphism and hyperlipidemia in Asians was not considered to exist. However, the T allele at this locus could be considered a risk factor for hyperlipidemia in Caucasians.

Meta-analysis of APOB EcorI (rs1042031)
This site was included in 10 case-control studies, involving a total of 3319 subjects, including 1633 in the hyperlipidemia group and 1686 in the control group. Baseline data and quality evaluation are shown in Table 4. Analysis of the association between A vs G alleles and hyperlipidemia (allele model) indicated heterogeneity (I 2 = 70.0%, P < 0.001), so the pooled effects were analyzed using a random-effects model. There was no significant difference in risk in individuals carrying the A or G alleles (OR = 1.333, 95% CI = 0.942-1.885, P = 0.104). The results of other gene models at this site were consistent with this conclusion, and so no association between the genetic polymorphism of APOB Ecor I and susceptibility to hyperlipidemia (Table 5) can be considered to exist.

Meta-analysis of APOE
This site was included in 30 case-control studies, involving a total of 11,614 subjects, including 5208 in the hyperlipidemia group and 6406 in the control group.
The baseline data and quality evaluation of the various studies are displayed in Table 7. The APOE ε3 allele was used as a reference to analyze the relationship between alleles and hyperlipidemia. Analysis of the data for ε2 (I 2 = 63.0%, P < 0.001) and ε4 (I 2 = 73.3%, P < 0.001) indicate that heterogeneity was present and so the pooled effects were analyzed using a random-effects model. The difference in risk between individuals with the ε2 and ε3 allele was not statistically significant (OR = 1.167, 95% CI = 0.955-1.426, P = 0.131). The risk of hyperlipidemia in individuals with the ε4 allele was higher than in those with the ε3 allele, a difference that was statistically significant (OR = 1.710, 95% CI = 1.405-2.083, P < 0.001) (Fig. 7). Because of heterogeneity, subgroup analysis by ethnicity was conducted, the results using the allele model demonstrating a risk of hyperlipidemia was different for Asians (OR = 1.304; 95% CI = 1.075-1.582, P = 0.007) for those with ε2 compared with the ε3 allele, but the association was not significant for Caucasians (OR = 0.807; 95% CI = 0.502-1.297, P = 0.376) (Fig. 8). There were significant differences in risk of hyperlipidemia, which was higher in both Asians (OR = 1.704; 95% CI = 1.325-2.192, P < 0.001) and Caucasians (OR = 1.759; 95% CI = 1.231-2.513, P = 0.002) with the ε4 allele than those carrying the ε3 allele (Fig. 9). Correlations in the APOE genotype (E2/E2, E2/E3, E2/ E4, E3/E4, E4/E4) and hyperlipidemia were analyzed using the wild type E3/E3 genotype as a reference. The heterogeneity, and OR and 95% CI values of these data are displayed in Table 8. The significance level was adjusted to α′ = α/(k-1) = 0.01. There was a significant difference in risk of hyperlipidemia between carriers of the E2/E4, E3/E4, and E4/E4 genotypes with carriers of the E3/E3 genotype, the P-values of which were < 0.01 in each case. To identify the source of significant heterogeneity, we conducted subgroup analysis based on ethnicity. The results demonstrated that there was a significant difference in risk of hyperlipidemia in carriers of all genotypes (E2/E2, E2/E3, E2/E4, E3/E4, E4/E4) compared with carriers of the E3/E3 genotype in Asians, while Caucasians carrying the E3/E4, E4/E4 genotypes were statistically different from those carrying E3/E3 (Table 9). Therefore, APOE gene polymorphisms can be considered to be closely associated with hyperlipidemia. For Asians, either the ε2 or ε4 allele was a risk factor for hyperlipidemia, while for Caucasians, only the ε4 allele was a risk factor.

Publication bias and sensitivity analysis
There was no apparent asymmetry in each Begg's funnel plot (Fig. 10), indicating that publication bias was slight. In addition, statistical analysis of the symmetry of Begg's funnel plots using an Egger's test demonstrated that publication bias for each gene locus displayed P-values all > 0.05, indicating that publication bias was apparently not present.
For groups that deviated substantially in the analysis, meta-analysis was performed again after exclusion of the associated manuscripts, and OR and P-values recalculated. Exclusion of the study [18] for APOA5-1131 T > C with the most deviating OR value using the allele model resulted in conclusions similar and consistent with those of the original data (OR = 1.800, 95% CI =   Exclusion of the manuscript [65] with the greatest deviation in data for the ε2 allele of APOE resulted in

Discussion
The present study found that allele C at APOA5-1131 T > C was a risk factor for hyperlipidemia, the A allele at AI-75 bp conferred susceptibility to hyperlipidemia, the T allele at APOB Xba I represents a preliminary pathogenic factor for hyperlipidemia in Caucasians, allele ε4 of the APOE gene is a risk factor for hyperlipidemia, and allele ε2 is a risk factor for hyperlipidemia in Asians.
The APOE gene, located on chromosome 19, contains 4 exons and 3 introns, with 3 isomers, and the functions by of regulating plasma total cholesterol (TC) and lipoprotein      3.96 0.14 T Kuusi [65] 1988 Helsinki, NR not reported, SRID single radial immunodiffusion metabolism. APOE3 is the most common phenotype. A principal function is to bind low-density lipoprotein receptor (LDL-R) and APOE receptor as the ligand [66]. Compared with APOE3, the ability of APOE4 to bind to its receptor is relatively strong, resulting in the metabolism of chylomicrons (CMs) and very low-density lipoprotein (VLDL) residues to be accelerated and the conversion of VLDL to LDL to increase. Additionally, the rate of liver internalization and catabolism of CM and VLDL residues becomes accelerated, resulting in increased free cholesterol in hepatocytes with feedback that caused a downregulation of LDL-R on their surface, resulting in a decrease in the metabolic rate of LDL [67]. Furthermore, the low intestinal cholesterol absorption capacity of ε4 carriers also increases, resulting in higher plasma levels of TC and LDL. This is consistent with the conclusion that the ε4 allele is a risk factor for hyperlipidemia in the present review. The study also found that the ε2 allele is harmful for blood lipid levels in the Asian population, but failed to establish the effects on blood lipid levels in the Caucasian population. This may be related to the imbalance of internal composition and the small sample size for Caucasians. Of course, we cannot rule out the possibility of a corresponding biological mechanism to explain why this locus has no harmful effects on Caucasians.
APOB is the principal protein component of LDL and plays a role in transportation of endogenous cholesterol to maintain its balance within the body. The APOB gene is located in region 23-24 of the short arm of human chromosome 2. The APOB gene plays a key role in the production, transport, and removal of LDL and VLDL from plasma and regulates the concentration of plasma cholesterol [68]. The polymorphism of the APOB XbaI restriction site is due to a mutation of nucleotide C → T at position 7673 of the APOB gene cDNA, which changes the codon sequence at position 2488 (ACC → ACT), thus producing an XbaI endonuclease recognition site. The T allele may be related to a reduction in LDL degradation rate mediated by the receptor [9]. A number of studies have also speculated that a single nucleotide polymorphism at this locus is a genetic marker and has linkage disequilibrium with other nearby DNA sequence variants that affect cholesterol levels [69]. Such a molecular mechanism could explain why the T allele is a risk factor for hyperlipidemia in Caucasians. Other studies further confirm our conclusions that this polymorphism of the APOB XbaI gene might increase the risk of cerebral infarction, and that the T allele is such a risk factor [70]. The T allele was associated with lower levels of HDL-C, which may be associated with the incidence of coronary heart disease [71]. The APOA1 gene is located in the terminal region of the long arm of chromosome 11 and consists of 3 introns and 4 exons. APOA1 is the main apolipoprotein to create highdensity lipoprotein (HDL), maintaining the stability and integrity of the HDL structure, and promoting the esterification of cholesterol (TC) [72]. The APOA1-75 bp polymorphism not only destroys the endonuclease recognition site but also changes the GGGCCGG sequence which activates transcription. A change in the sequence may also affect the synthesis of APOA1 [73]. This mechanism is consistent with the conclusion that there is an association between the A1-75bp gene single nucleotide polymorphisms and hyperlipidemia. The APOA5 gene, located in 23 regions of the long arm of chromosome 11, has 1889 bps and consists of 4 exons, 2 introns, and 4 silencing molecules. APOA5 can reduce triglyceride (TG) and increase HDL, representing a protective factor for coronary heart disease [74]. Some of the manuscripts also clearly stated that the mutation APOA5-1131 T > C is closely related to increased triglyceride levels [75] and that the CC genotype of this locus was positively correlated with serum TG levels and negatively correlated with APOA5 levels [76].
A meta-analysis can effectively compensate for the lack of statistical efficacy and other problems within a single study. However, although the present review developed a scientifically-based and comprehensive search strategy with strict unified screening criteria, limitations still remain [77]: (1) There were few relevant Chinese and English manuscripts on the acquisition of particular gene loci, such as APOAI and APOB MspI, so the number of case-control studies included in the analysis was small, possibly reducing the effectiveness of the Egger's and Begg's tests, in addition to sensitivity analysis; (2) The data included in the review did not involve additional races, which led to heterogeneity. Although ethnic subgroup analysis can identify heterogeneity to some extent, we found that there was a small sample size in Caucasians for APOB XbaI, possibly the reason why the results of the genetic model were not consistent at this locus.
(3) It is unknown whether there were statistical differences in sex and age among individuals included in the study; (4) The effects of gene-environmental interactions and genetic linkage disequilibrium were not considered. In the future, we shall include more reliable data in this

Conclusions
In summary, the results of the present meta-analysis revealed that the C allele of APOA5 1131 T > C, the A allele at APOA1-75 bp, the APOB XbaI T allele, and the ε2 and ε4 alleles of APOE may represent genetic risk factors for susceptibility for hyperlipidemia. In addition, we found it is consistent with the present study on the pathological mechanisms of hyperlipidemia. However, there is a need for further large-scale studies, including larger case-control studies and analysis of other loci of the APO genes, to confirm our conclusions and elucidate the influence of gene-environment interactions.

Identification of studies for inclusion
The inclusion criteria for the present meta-analysis were as follows: (1) studies that evaluated the association between APO and risk of hyperlipidemia; (2) studies with an appropriate statistical design and selection methods; (3) case-control and RCT studies; (4) diagnostic criteria for dyslipidemia that were clear and uniform [79]; (5) distribution of APO genotypes in controls group were consistent with the Hardy-Weinberg equilibrium (HWE); (6) allele typing methods were accurate; (7) data included in the studies were complete, without omissions. Duplicated data, reviews, abstracts, case reports, animal studies, and studies that did not meet the inclusion criteria were excluded.

Data extraction
Two reviewers (XNZ and QS) independently conducted literature screening and evaluation. The following information was extracted from each study for inclusion in the review: first author, year of publication, area, age, source of control, sample size of controls and cases, genotyping method, Hardy-Weinberg equilibrium (HWE), and the distribution of genotypes and frequencies of alleles in cases and controls. Any disputes were resolved by discussion with a third investigator.

Quality evaluation
The quality of the selected case-control studies was evaluated according to the Newcastle-Ottawa Quality Assessment Scale (NOS) [80], of which data with scores  10 Begg's funnel plot for the APOE ε4 allele 0-3, 4-6 or 7-9 were low, moderate or high-quality, respectively [81].

Statistical analyses
The included hyperlipidemia data were analyzed by meta-analysis using Stata 11 software. The correlation between apolipoprotein gene polymorphism and hyperlipidemia was expressed by odds ratio (OR) and 95% confidence intervals (CIs). In order to better evaluate the presence of heterogeneity between the studies, an I 2 test was also used. Where homogeneity (I 2 < 50%) was identified in the meta-analysis, a fixed-effects model was adopted; otherwise, a random-effects model was used to integrate the incorporated data. The data were assessed using Egger's and Begg's tests to evaluate publication bias. Sensitivity analysis was conducted by deleting, in turn, the data from individual studies that had large deviations as identified in the results, then recalculating the OR value. All P-values were two-sided, with a significance threshold set at α = 0.05.
To explore the source of significant heterogeneity, subgroup analysis of race was performed. A total of 7 sites were included, of which 3 sites (APOA5-1131 T > C,APOB XbaI, and APOE) were evaluated by subgroup analysis of ethnicity, 2 sites (APOB MspI, and APOB EcorI) were analyzed by sensitivity analysis, as there was only one published study of different races in the literature that was not suitable for subgroup analysis. Race was not evaluated in 2 sites (APOA1-75 bp, APOA1 + 83 bp) by subgroup analysis due to the fact that the populations studied were the same race, and had no significant heterogeneity.