Meta-analysis for the Association of Apolipoprotein E ε2/ε3/ε4 Polymorphism with Coronary Heart Disease.

BACKGROUND: Coronary heart disease (CHD) is a multifactorial disease and is thought to have a polygenic basis. Apolipoprotein E (APOE) gene is one such candidate with its common ε2/ε3/ε4 polymorphism in CHD. In recent years, numerous case-control studies have investigated the relationship of APOE polymorphism with CHD risk. However, the results are confusing.
METHODS: To clarify this point, we undertook a meta-analysis based on 14 published studies including 5746 CHD cases and 19,120 controls. Crude odds ratios (ORs) with 95% confidence intervals (CIs) were assessed for association using a random-effects or fixed-effects model using STATA version 10 (StataCorp LP, College Station, TX, USA).
RESULTS: Overall, the analysis showed that carriers of APOE ε2 allele decreased risk for CHD (ε2 allele vs. ε3 allele: OR = 0.82, 95% CI: 0.75-0.90, P < 0.001; ε2 carriers vs. ε3 carriers: OR = 0.81, 95% CI: 0.73-0.89, P < 0.001), compared with those carrying ε3 allele, especially in Caucasian population. However, those with ε4 allele had a significant increased risk for CHD (ε4 allele vs. ε3 allele: OR = 1.34, 95% CI: 1.15-1.57, P < 0.001), especially in Mongoloid population. Potential publication bias was observed in the genetic model of ε4 versus ε3, but the results might not be affected deeply by the publication bias. When we accounted for publication bias using the trim and fill method, the results were not materially alerted, suggesting the stability of our results.
CONCLUSIONS: Taken together, our meta-analysis supported a genetic association between APOE gene and CHD. ε4 increased the risk of CHD, whereas ε2 decreased the risk of CHD.

INTRODUCTION

Coronary heart disease (CHD), a complex disease, is the most important cause of mortality and morbidity worldwide. CHD accounts for 7.3 million death in 2008.[1] CHD is often initially detected from clinical manifestations such as myocardial infarction, angina, or sudden death due to artery occlusion.

The exact etiology underlying CHD is still unclear. Many environmental and genetic factors are identified to implicate the development of CHD. The environmental factors, such as tobacco smoking, physical activity and alcohol intake, and dietary fat consumption, in particular, the amount of saturated fatty acids, have been considered as the main risk factors for CHD.[2] In recent years, several susceptibility genes have been proposed to manifest an association with CHD.

Among the genetic determinants that have been more widely studied related to CHD risk is the common polymorphism in the apolipoprotein E (APOE) gene.[3] APOE is located on chromosome 19q13.2, with 3 common isoforms (ε2, ε3 and ε4) and 6 genotypes (E2E2, E2E3, E3E3, E2E4, E3E4, and E4E4).[4] The alleles differ at two single-nucleotide polymorphisms, giving rise to different amino acids at protein positions 112 and 158; cysteine at 112 and arginine at 158 for ε3, two cysteine for ε2 and two arginine for ε4. ε3 accounting for approximately 50–90% in all population is considered to be the wild type allele whereas alleles ε2 and ε4 are considered variants which vary between 5-35% and 1-15%, respectively.[56] APOE is highly polymorphic and plays an imperative role in endogenous lipoprotein metabolism and tissue distribution. It is present in lipoprotein particles and mediates lipoprotein through binding to the low-density lipoprotein (LDL) and lipoprotein remnant receptors. Defects in the APOE protein could diminish its ability to bind to the receptors, which may lead to an elevated blood cholesterol level.[7] It has been well-established that the presence of the APOE ε4 allele is associated with higher levels of total and LDL serum cholesterol, whereas the presence of the APOE ε2 allele is associated with the lower effect with reference to cholesterol effects from the ε3 allele.[8] These results suggest that APOE is associated with CHD.

To date, numerous epidemiological studies[29101112131415161718192021222324] have recently focused on the association between APOE ε2/ε3/ε4 polymorphism and CHD risk. However, results in different studies are inconsistent. For example, Singh et al. found a significant association of ε4 polymorphism, especially E3/E4 genotype with CHD, along with high-density lipoprotein (HDL) and LDL concentrations. However, Marrzoq et al. suggested APOE genotypes were not associated with CHD. Moreover, there was no significant difference between the mean of triglyceride and HDL levels among different APOE genotypes. Therefore, the present meta-analysis is designed to derive a more precise estimation.

METHODS

Identification of studies

Relevant studies were identified from the following electronic databases: PubMed, EMBASE, Web of Science, CNKI, CBM, Wanfang, VIP and hand searching of relevant journals and the reference lists of retrieved articles, with the last updated search performed before September 2012. The search was performed using various combinations of keywords including (“APOE”) and (“polymorphism” or “variant” or “mutation”) and (“CHD”) without language limitation. The references of the selected papers were also checked by hand-search for other potential articles that possibly have been missed in the initial search. To be eligible for inclusion in this meta-analysis, a study must meet the following criteria: (1) APOE gene polymorphism in CHD; (2) sufficient data for calculating an odds ratio (OR) with its 95% confidence interval (CI). When duplicate articles were published, only the newest or most informative single article was selected. Study was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement.[25]

Data extraction

All the data were extracted independently by two reviewers with the standard protocol. Potential disagreements were documented and resolved by discussion with a third reviewer. The following characteristics were extracted from each study: First author, year of publication, ethnicity of the study population, sample size, genotype method; source of case and control subjects.

Quality assessment

The quality of the included studies was assessed using the following criteria modified from the previous report, and studies with four or five adequate items were defined as high quality.

Description of the case and control groups (not described, inadequate, adequate).

Assessment and validation of CHD in the patients (not described, inadequate, adequate). Adequate validation would include confirmation by coronary angiography; inadequate validation would include recollection of the patient as the only clinical evidence without examination.

Description of the laboratory procedures for the genotyping (not described, inadequate, adequate).

Confirmation of Hardy-Weinberg equilibrium (HWE) in the genotype distribution (not described, inadequate, adequate). Since HWE is a surrogate to assess study quality, and the effect of HWE is associated with problems in the design and conduct of genetic association studies, studies with departures from HWE were defined as low-quality studies.

Equal assessment for confounding factors in the case and control groups (not described, inadequate, adequate).

Statistical analyses

Allele frequencies at genetic polymorphisms from each study were determined by the allele counting method. HWE for the controls in each study was assessed by Chi-square interval or Fisher exact test. The combined OR and 95% CI was used to compare contrasts of genotypes and alleles between cases and controls. Statistical heterogeneity was measured using the Chi-square based Q statistic (P < 0.10 was considered statistically significant heterogeneity). We also quantified the effect of heterogeneity using I2 statistic:[26]

I2=100% × (Q – df)/Q

I2 values of 25%, 50%, and 75% were defined as low, moderate, and high estimates, respectively. Either a random-effects model (DerSimonian-Laird method)[27] or fixed-effects model (Mantel-Haenszel method)[28] was used to calculate pooled effect estimates in the presence or absence of heterogeneity, respectively.

Finally, potential publication bias was evaluated through funnel plot visual analysis and with the Begg's and Egger's tests.[2930] A P < 0.05 was considered statistically significant. For the possible publication bias, we used trim and fill method to evaluate the influence to the results. All statistical analyses were performed by STATA version 10 (StataCorp LP, College Station, TX, USA).

RESULTS

Studies included in the meta-analysis

The flow chart of literatures selection was provided in Figure 1. There were 161 papers relevant to the search words. Through screening the title and reading the abstract, 135 papers were excluded. Of these, the type of 26 papers was review. Twenty-two papers were not related to the genetic polymorphism. Twenty-seven papers were not about APOE gene polymorphism. Fifty-four papers were not about CHD. Six papers were published in other language (Russian: 2; Spanish 2; Portuguese: 1; Norwegian: 1). Twenty-six studies were left for full publication review. Of the 26 studies, 2 studies did not have a control group, 5 did not provide enough information on genotype distribution, leaving 19 studies for more detailed assessment. Two studies[3132] were excluded because of overlap data. Four studies[13151924] excluded were deviated from HWE. One study[17] contained data on two different groups (Whites and African Americans). Thus, in accordance with the objectives of the present study, we analyzed them independently. As a result, a total of 14 studies from 13 independent articles including 5746 CHD cases and 19,120 controls were subjected to the final analysis. The gene discussed in all included studies was consistent with target gene. Of the 14 studies, 6 studies were conducted in Mongoloid population, 5 were conducted in Caucasians population, and 3 in other population. Characteristics of studies included in the current meta-analysis are shown in Table 1.

Figure 1

The flow diagram of study selection.

Table 1

Basic information of studies included in the meta-analysis

Author	Year	Population	Sample size (n)		Characteristics of the included studies		Genotype method	HWE for controls (P)	Case source	Control source	Quality assessment
			Cases	Controls	CHD group	Control group
Ou et al.	1998	Japanese	214	310	86.5% male, age: 55.4 ± 7.4 years BMI: 23.2 ± 2.6	86% male, Age: 54.5 ± 5.9 years BMI: 23.9 ± 2.8	PCR-RFLP	0.095	Hospital (A hospital in Tsuchiura city)	Hospital (A hospital in Tsuchiura city)	High
Cao et al.	1999	Chinese	78	85	NA	NA	PCR-RFLP	0.789	Hospital (Han population in northeast China)	NA (Han population in northeast China)	Low
Yang et al.	2001	Chinese	204	136	CHD1 age: 45.4 ± 6.8 years, BMI: 23.8 ± 2.6 CHD2 age: 70.1 ± 6.4 years, BMI: 24.2 ± 2.3	Age: 52.5 ± 12.8 years BMI: 34.1 ± 2.4	PCR-RFLP	0.491	Hospital	Population (A local company)	High
Freitas et al.	2002	Caucasian	640	624	Age: 44 ± 4 years 88% Male	Age: 40 ± 6 years 52% Male	PCR-RFLP	0.767	Hospital	Population (local community)	High
Wang et al.	2004	Chinese	186	350	Age: 65.0 ± 10.6 years	Age: 63.6 ± 8.3 years	Multi- ARMS	0.431	Hospital (Han population in Hubei)	Hospital (Han population in Hubei)	High
Ouyang et al.	2005	Chinese	200	100	Male/female: 105/95 Age: 63.6 ± 5.8 years BMI: 15.1 - 34.3kg/m2	Male/female: 55/45 Age: 62.1 ± 7.8 years	PCR-RFLP	0.399	Hospital (Han population in Beijing and Hebei)	Volunteers (Han population in Beijing and Hebei)	High
Volcik et al.	2006	Whites/African Americans	1017/534	8287/5840	NA	NA	TaqMan	0.807/0.235	Population (four US communities)	Population (four US communities)	High
Akanji et al.	2007	Kuwait	50	65	Age: 54.0 (40.0-76.0) years BMI: 28.1 (22.5-38.1)	Age: 39.0 (25.0-60.0) years BMI: 29.8 (21.0-41.7)	PCR-RFLP	0.441	Hospital (Kuwaiti Arab)	Volunteers (Kuwaiti Arab)	High
Singh et al.	2008	Indians	193	150	Age: 54.94 ± 11.43 years Male/female: 74.6%/25.4%	Age: 53.42 ± 12.47 years Male/female: 70%/30%	PCR	0.327	Hospital (northwest India)	Volunteers (northwest India)	High
Kolovou et al.	2009	Athens	359	248	Normoweight (98): Age: 61 ± 11 years, BMI: 24 ± 1 Overweight (189): Age: 59 ± 10 years, BMI: 27 ± 1 Obese (72): Age: 59 ± 10 years, BMI: 32 ± 2	Age: 30 ± 12 years BMI: 22 ± 3	PCR-RFLP	0.133	Hospital (Greek Caucasian)	Volunteers (Greek Caucasian)	High
Marrzoq et al.	2011	Asians	69	68	Male/female: 45/24	Male/female: 33/35	PCR-RFLP	0.708	Hospital (Gaza Strip)	NA (Gaza Strip)	High
Corella et al.	2011	Spanish	534	1123	Male/female: 424/110 Age: 54.0 ± 7.3 years BMI: 29.2 ± 3.5	Male/female: 884/239 Age: 53.7 ± 7.2 years BMI: 28.7 ± 3.7	Taqman	0.169	Volunteers (Spanish EPIC cohort)	Volunteers (Spanish EPIC cohort)	High
Gustavsson et al.	2012	Sweden	1735	4654	NA	2642 women	PCR-SPM	0.766	Population (INTERGENE and SHEEP)	Population (INTERGENE and SHEEP)	High

NA: Not available; HWE: Hardy-Weinberg equilibrium; PCR-RFLP: Polymerase chain reaction with restriction fragment length polymorphism; PCR-SPM: Polymerase chain reaction with solid-phase minisequencing; BMI: Body mass index; CHD: Coronary heart disease.

According to the quality criteria, there were thirteen studies with high quality, and only one with low quality.

Main results of meta-analysis

A summary of meta-analysis findings regarding the relation between the APOE ε2/ε3/ε4 polymorphism is presented in Table 2. Forest plots are shown in Figures 2–5.

Table 2

Summary estimates of the association between APOE gene polymorphism and CHD: Overall analysis and subgroup analysis

Polymorphism	Subgroup	Number of studies	Test of association			Test of heterogeneity
			OR	95% CI	PZa	Modelb	I2 (%)	PQc
ε2 allele versus ε3 allele	All	14	0.82	0.75–0.90	<0.001	F	27.1	0.164
	High quality studies	13	0.84	0.77–0.91	<0.001	F	33.9	0.111
	Race
	Mongoloid	6	1.07	0.83–1.37	0.611	F	0.0	0.848
	Caucasian	5	0.71	0.63–0.81	<0.001	F	0.0	0.893
	Others	3	0.96	0.84–1.09	0.488	F	0.0	0.739
ε4 allele versus ε3 allele	All	14	1.34	1.15–1.57	<0.001	R	71.3	<0.001
	High quality studies	13	1.28	1.12–1.46	<0.001	R	73.1	<0.001
	Race
	Mongoloid	6	1.97	1.61–2.41	<0.001	F	0.0	0.951
	Caucasian	5	1.04	0.97–1.12	0.271	F	0.0	0.939
	Others	4	1.11	1.02–1.21	0.018	R	77.7	0.011
ε2 carriers versus ε3 carriers	All	14	0.81	0.73–0.89	<0.001	F	12.8	0.313
	High quality studies	13	0.83	0.76–0.90	<0.001	F	21.8	0.223
	Race
	Mongoloid	6	1.01	0.79–1.29	0.927	F	0.0	0.877
	Caucasian	5	0.72	0.64–0.81	<0.001	F	0.0	0.838
	Others	3	0.94	0.82–1.07	0.348	F	0.0	0.630
ε4 carriers versus ε3 carriers	All	14	1.29	1.12–1.48	<0.001	R	60.0	0.002
	High quality studies	13	1.21	1.08–1.35	0.0001	R	62.3	0.001
	Race
	Mongoloid	6	1.76	1.44–2.15	<0.001	F	0.0	0.933
	Caucasian	5	1.05	0.98–1.13	0.147	F	0.0	0.978
	Others	3	1.08	0.99–1.18	0.065	R	73.1	0.024

aZ test used to determine the significance of the overall OR. PZ < 0.05 was considered to be significant; bF: Fixed-effects model; R: Random-effects model; cChi-square Q statistic test used to assess the heterogeneity in each group. PQ > 0.1, Fixed-effects model was used, PQ < 0.1, Random-effects model was used. OR: Odds ratio; CI: Confidence interval; CHD: Coronary heart disease; APOE: Apolipoprotein E.

Figure 2

Forest plot on the association between ε2 allele (vs. ε3 allele) and CHD.

Figure 5

Forest plot on the association between ε4 carriers (vs. ε3 carriers) and CHD.

Forest plot on the association between ε2 carriers (vs. ε3 carriers) and CHD.

Forest plot on the association between ε4 allele (vs. ε3 allele) and CHD.

Meta-analysis showed a decreased risk for the ε2 allele compared with the ε3 allele in overall population (OR = 0.82, 95% CI: 0.75–0.90, P < 0.001), without heterogeneity among studies (I2 = 27.1%, P = 0.164). When stratified by race, the pooled OR was just statistically significant in Caucasian population, not in Mongoloid and others. We also compared the model of ε2 carriers versus ε3 carriers. Under a fixed-effect model, the pooled OR for ε2 carriers versus ε3 carriers was 0.81 (95% CI: 0.73–0.89, P < 0.001). Analysis after stratification by race indicated significant association with CHD in Caucasian population. Therefore, carriers of APOE ε2 allele had a decreased risk for CHD, especially in Caucasians.

Meta-analysis showed an increased risk for the ε4 allele compared with the ε3 allele in overall population (OR = 1.34, 95% CI: 1.15–1.57, P < 0.001) with significant inter-study heterogeneity (I2 = 71.3%, P < 0.001). When stratified by race, heterogeneity was not detected. The combined OR obtained using a fixed-effect model was significant in Mongoloid population. Caucasians failed to find an association. Similarly, for genotype contrast of ε4 carriers versus ε3 carriers, a significant association was noticed in overall population (OR = 1.29, 95% CI: 1.12–1.48, P < 0.001). Analysis after stratification by race indicated a significant association with CHD risk in Mongoloid population (OR = 1.76, 95% CI: 1.44–2.15, P < 0.001). Therefore, carriers of APOE ε4 allele had an increased risk for CHD, especially in Mongoloid population.

Publication bias

Both Begg's test and Egger's test were performed to assess the publication bias of the literature. Funnel plots on the distribution of the ORs from individual studies in relation to their respective standard deviation are shown in Figures 6 and 7. No evidence of publication bias was observed in the genetic model of overall ε2 allele versus ε3 allele (PBegg = 0.583, PEgger = 0.820). However, the shape of the funnel plot for the genetic model of overall ε4 allele versus ε3 allele seemed to be asymmetry, suggesting potential publication bias (PBegg = 0.002). This might be a limitation for our analysis because the studies with null findings, especially those with small sample size, were less likely to be published. By using the trim and fill method, we showed that if the publication bias was the only source of the funnel plot asymmetry, it needed seven more studies to balance the funnel plot. The adjusted risk estimate was attenuated but remained significant (ε4 allele vs. ε3 allele: OR = 1.08, 95% CI: 1.02–1.15, P = 0.009; ε4 carriers vs. ε3 carriers: OR = 1.08, 95% CI: 1.02–1.16, P = 0.016), indicating the stability of our results.

Figure 6

Funnel plot for the results of the meta-analysis of the ε2 allele vs. ε3 allele.

Figure 7

Funnel plot for the results of the meta-analysis of the ε4 allele vs. ε3 allele.

DISCUSSION

Many epidemiological studies have investigated the relationship between the APOE polymorphism and the risk of CHD, but because of small sample size and the low statistical power of individual studies, results have been contradictory. The main purpose of the present meta-analysis is to summarize the published literature and attempt to clarify the role of APOE ε2/ε3/ε4 polymorphism in CHD. We quantitatively combined 18 case–control studies including 6342 CHD cases and 19,688 controls, and found that carrying one or more ε4 allele was associated with an increased risk for CHD, whereas individuals with one or more ε2 allele had a significantly lower risk of developing CHD compared to those with ε3 allele. Therefore, it is safe to assume that ε4 allele is an independent factor for the development of CHD, however, ε2 allele tends to be a protective factor.

We also conducted subgroup analysis by race. As for the Caucasian population, significant association was observed in the contrast of the genetic model of ε2 allele vs. ε3 allele and ε2 carriers vs. ε3carriers. However, the genetic model ε4 allele versus ε3 allele and ε4 carriers versus ε3carriers remained significant difference only in Mongoloid population. Therefore, ethnic difference may contribute to this phenomenon.

Utermann et al.[33] first reported that the presence of APOE polymorphisms has inspired widespread interest in the genetic associations between these single-nucleotide polymorphisms and many complex diseases including Alzheimer's disease,[34] type 2 diabetes,[35] hypertension,[36] and others. Besides, several studies have provided evidence that APOE is functional. Structural defects in APOE could result in an impaired interaction between APOE-containing lipoproteins and their receptors and induce the development of atherogenic dyslipidemias and premature cardiovascular disease, including CHD. Several former meta-analyses had documented that ε4 was associated with increased risk for CHD.[3738] In our study, we also found that ε2 carriers had a lower risk of CHD.

It is worth mentioning that testing for deviations from HWE in controls is an important requirement in population genetic studies. HWE is based on five basic assumptions: (1) The population is large (i.e., there is no genetic drift); (2) there is no gene flow between population due to migration or transfer of gametes; (3) mutations are negligible; (4) individuals are mating randomly; and (5) natural selection is not operating on the population. Deviation from HWE may point to genotyping error, racial heterogeneity, or selection bias. In our study, four of 18 studies were not conformed to HWE. We excluded the four studies in combined analysis.

Heterogeneity did not exist in overall comparisons of ε2 allele versus ε3 allele and ε2 carriers versus ε3carriers, but in the genetic model of ε4 allele versus ε3 allele and ε4 carriers versus ε3carriers, which may affect the results of the present meta-analysis. We adopted the random-effects model, as it is more conservative than the fixed-effects model and accounts for additional sources of inter-study variation when heterogeneity exists. This heterogeneity may attribute to the potential confounding due to case definition and sampling, sample size, genotyping method, difference of ethnicity, or it may be due to interaction with other risk factors. In order to understand the sources of heterogeneity, we conducted the subgroup analysis by race. The heterogeneity disappears in each subgroup. Hence, the heterogeneity might be resulted from the ethnicity difference.

We had to state the potential publication bias. Significant bias was observed in genetic model of ε4 allele versus ε3 allele and ε4 carriers versus ε3 carriers. The explanations might arise from some aspects. First, our meta-analysis took into consideration only fully published data, and the abstract and conference papers were excluded. Second, this meta-analysis only focused on papers published in English language, and some eligible studies which were reported in other languages might be missed. Third, positive results tend to be accepted by journals while negative results are often rejected or not even submitted. We should point out that the publication bias might partly account for the results, but which were not affected deeply. When we adjusted the results using the trim and fill method, the adjusted risk estimate was attenuated but remained significant, indicating the stability of our results. Hence, the summary statistics obtained may approximate the actual average.

In addition, heterogeneity and publication bias, some limitations of this meta-analysis should be acknowledged. Due to the relative small number of the eligible studies, we were unable to perform further subgroup analysis such as by gender, body mass index. Besides, CHD is a complex disease with various involved factors including environmental and genetic factors. However, many eligible studies included in this meta-analysis did not consider most of the important environmental factors, moreover, a single gene polymorphism is unlikely to provide the complete explanation of genetic risk for CHD. Therefore, gene-gene and gene-environment interactions should be considered in future studies.

Our study suggested a genetic association between APOE gene and CHD, ε4 increased the risk of CHD, whereas ε2 decreased the risk of CHD. Well-designed case-control studies with larger sample sizes regarding the association of APOE polymorphism and CHD need to be performed in the future.