Influence of ATP-Binding Cassette Transporter 1 R219K and M883I Polymorphisms on Development of Atherosclerosis: A Meta-Analysis of 58 Studies

Background Numerous epidemiological studies have evaluated the associations between ATP-binding cassette transporter 1 (ABCA1) R219K (rs2230806) and M883I (rs4149313) polymorphisms and atherosclerosis (AS), but results remain controversial. The purpose of the present study is to investigate whether these two polymorphisms facilitate the susceptibility to AS using a meta-analysis. Methods PubMed, Embase, Web of Science, Medline, Cochrane database, Clinicaltrials.gov, Current Controlled Trials, Chinese Clinical Trial Registry, CBMdisc, CNKI, Google Scholar and Baidu Library were searched to get the genetic association studies. All statistical analyses were done with Stata 11.0. Results Forty-seven articles involving 58 studies were included in the final meta-analysis. For the ABCA1 R219K polymorphism, 42 studies involving 12,551 AS cases and 19,548 controls were combined showing significant association between this variant and AS risk (for K allele vs. R allele: OR = 0.77, 95% CI = 0.71–0.84, P<0.01; for K/K vs. R/R: OR = 0.60, 95% CI = 0.51–0.71, P<0.01; for K/K vs. R/K+R/R: OR = 0.69, 95% CI = 0.60–0.80, P<0.01; for K/K+R/K vs. R/R: OR = 0.74, 95% CI = 0.66–0.83, P<0.01). For the ABCA1 M883I polymorphism, 16 studies involving 4,224 AS cases and 3,462 controls were combined. There was also significant association between the variant and AS risk (for I allele vs. M allele: OR = 0.85, 95% CI = 0.77–0.95, P<0.01). Conclusions The present meta-analysis suggested that the ABCA1 R219K and M883I polymorphisms were associated with the susceptibility to AS. However, due to the high heterogeneity in the meta-analysis, the results should be interpreted with caution.


Introduction
Atherosclerosis (AS) is a common, multifactorial disease that causes significant morbidity and mortality worldwide [1]. Its exact mechanisms are still unclear. Multiple environmental and lifestyle factors may significantly contribute to the AS risk, such as climate, diet, age, and economic status. The risk of developing AS tends to run in families, which suggests the genetic factors play a crucial role in the development of AS. Until recently, many studies have focused on this field, and ATP-binding cassette transporter 1 (ABCA1) gene has been extensively studied .
It is widely accepted that a low level of plasma high-density lipoprotein cholesterol (HDL-C) is a major risk factor for atherosclerotic diseases in the general population [49]. ABCA1, a member of the ATP-binding cassette family,could promote the HDL formation and the cholesterol efflux, thus impacting the development of AS [50]. Previous studies have reported that the mutations in ABCA1 are responsible for two forms of heritable HDL disorders, namely Tangier disease (Homozygosity) and familial hypoalphalipoproteinemia (heterozygosity) [51,52]. Given the crucial role of ABCA1 in the HDL formation and the cholesterol efflux, the mutations in ABCA1 may also play a significant role in the development of atherosclerotic diseases. Recently, a number of molecular epidemiological studies have been done to evaluate the associations between the ABCA1 gene polymorphisms (such as R219K, M883I, C69T, V825I, R1587K, V771M and 2565C/T) and the risk of atherosclerotic diseases . However, results of different studies have been inconsistent, possibly due to small sample sizes in the individual studies. In

Inclusion and Exclusion Criteria
To be included in the present meta-analysis, the studies had to comply with the following major criteria: (1) case-control or cohort studies evaluating the associations between the ABCA1 R219K and M883I polymorphisms and AS risk; (2) published studies with full text articles; (3) Diagnosises of atherosclerotic diseases were made according to the internationally recognized diagnostic criterion. The diagnosis of ischemic heart disease (CAD and MI) is accorded with the result of coronary angiography, criteria of World Health Organization (WHO), criteria of European Society of Cardiology (ESC), or criteria of American College of Cardiology/American Heart Association (ACC/AHA), and the diagnosis of IS is accorded with result of computed tomography (CT) or magnetic resonance imaging (MRI); (4) sufficient published data for calculating odds ratios (ORs) with their 95% confidence intervals (CIs).

Data Extraction
Data, including name of the first author, year of publication, study population (country, ethnicity), study type (case-control and cohort study), type of atherosclerotic disease, source of controls (population-based studies and hospital-based studies), sample size (total numbers of cases and controls), and number of genotypes in cases and controls, were extracted from each study by two reviewers independently (Yin YW and Zhang LL) according to the pre-specified inclusion criteria. Decisions were compared and disagreements about study selection were resolved by consensus or by involving a third reviewer (Li JC).

Quality Assessment for Individual Studies
The quality of the individual studies was evaluated and scored by two reviewers independently based on the Newcastle-Ottawa Scale (NOS) [57]. Each study was assessed based on three broad perspectives: selection, comparability, and exposure, and each satisfactory answer received one point. The NOS ranges between zero (none of the quality criterion was met) up to nine stars (all the quality criteria were met), and the high-quality study was considered as the one with a score equal to or higher than seven. The third reviewer (Li BH) examined the results, and a consensus was reached.

GRADE Quality Assessment
GRADE (Grades of Recommendation, Assessment, Development and Evaluation) approach was adopted to grade quality of evidence for each association [58]. The GRADE system included: level of evidence: (1) high quality; we are very confident that the true effect lies close to that of the estimate of the effect, (2) moderate quality; we are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different, (3) low quality; our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect and (4) very low quality; we have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect. Two reviewers (Wang JZ and Zhang MJ) assessed quality independently and solved disagreement by discussion.

Statistical Analysis
For the controls of each study, Hardy-Weinberg equilibrium (HWE) was assessed using the chi-square test (P,0.05 was considered significant deviation from HWE  [59,60]. Meta-regression was performed subsequently to explore the heterogeneity sources. The pooled OR was estimated by the random-effects model. Subgroup analyses were performed based on ethnicity (Caucasians and Asians), atherosclerotic diseases (CAD and IS), source of controls (population-based studies and hospital-based studies) and study type (case-control and cohort study). Sensitivity analyses were performed based on HWE (studies without HWE were excluded), NOS score (studies with score #6 were excluded) and article type (master theses were excluded). Begg's funnel plot and Egger's regression test were conducted to identify possible publication bias in the current meta-analysis (P,0.05 was considered representative of statistically significant publication bias) [61]. In addition, fail-safe number (Nfs 0.05 ) was calculated to explore the extent to which the ''file drawer problem'' may have affected study results. All statistical analyses were done with Stata software (Version 11; StataCorp LP, College Station, TX).

Characteristics of the Studies
The present study met the PRISMA statements and MOOSE guidelines (Checklist S1, Table S1 and Fig. 1). A total of 1184 articles were identified after searching. After careful review, 47 articles involving 58 studies (42 studies for R219K polymorphism and 16 studies for M883I polymorphism) met the inclusion criteria and were selected in this meta-analysis . For the ABCA1 R219K polymorphism, 12551 AS cases and 19548 controls were included to assess the association between the variant and AS risk . For the ABCA1 M883I polymorphism, 4224 AS cases and 3462 controls were identified to assess the association of the variant with AS risk [14,15,22,23,29,30,[42][43][44][45][46][47][48]. Main characteristics of the included studies were listed in Table 1 and Table 2. The most commonly atherosclerotic disease included in the present meta-analysis was CAD. In addition, there were nine studies involving IS, five studies involving MI, and one study involving IS and carotid artery atherosclerosis (CAA). There were 40 studies of Asians, 18 studies of Caucasians, and one study of mixed population. Four studies did not follow the HWE [16,18,42,46] and two studies have insufficient data for calculation of the HWE [22,29]. The NOS results showed that the average scores were 6.9 and 6.8, respectively. In addition, the results of GRADE were shown in Table S2.

Sensitivity Analysis
Sensitivity analyses were conducted to determine whether modification of the inclusion criteria of the meta-analysis affected the final results. The included studies were limited to those conforming to HWE or the high quality studies (NOS score $7). In addition, we also performed sensitivity analysis by removing the master thesis. Overall, the corresponding pooled ORs were not materially altered, either for the ABCA1 R219K polymorphism or for M883I polymorphism. This suggested that the overall results of this meta-analysis were statistically robust. The main results of sensitivity analyses were shown in Table 2.
For the ABCA1 R219K polymorphism, heterogeneity can be explained by the subtype of atherosclerotic diseases (allelic model: P T2 = 0.013, additive model: P T2 = 0.034, and recessive model: P T2 = 0.017) and study type (dominant model: P T4 = 0.037). According to the results of the meta-regression, we performed subgroup analyses based on the atherosclerotic diseases (CAD and IS) and study type (case-control and cohort study). Heterogeneity was distinctly reduced, although it was still significant in the allelic model (IS group: P Q = 0.0005, I 2 = 71%; CAD group: P Q ,0.01, I 2 = 79%), additive model (IS group: P Q = 0.001, I 2 = 69%; CAD group: P Q ,0.01, I 2 = 68%), recessive model (IS group: P Q = 0.0008, I 2 = 70%; CAD group: P Q ,0.01, I 2 = 59%) and dominant model (CS group: P Q = 0.14, I 2 = 42%; CCS group: P Q ,0.01, I 2 = 66%). For the ABCA1 M883I polymorphism, we did not indentify the sources of heterogeneity by the metaregression. The main results of meta-regression were shown in Table S3 and Table S4.

Publication Bias
Begg's funnel plot and Egger's regression test were performed to assess potential publication bias. For the ABCA1 R219K polymorphism, visual inspection of the funnel plot ( Fig. 4: K allele vs. R allele) displays asymmetrical distribution of OR estimations, suggesting significant publication bias. In addition, the results of Egger's regression test also provided evidence for publication bias (P,0.01 for allelic model, P,0.01 for additive model, P,0.01 for recessive model, and P,0.01 for dominant model). For the ABCA1 M883I polymorphism, no obvious asymmetry was observed in any genetic model according to the visual assessment of funnel plot ( Fig. 4: I allele vs. M allele). Moreover, the results of Egger's regression test did not provide any statistical evidence for publication bias (P = 0.34 for allelic model, P = 0.24 for additive model, P = 0.10 for recessive model, and P = 0.09 for dominant model). In addition to the above analyses, we also calculated the Nfs (for the ABCA1 R219K polymorphism, Nfs = 133 for allelic model, Nfs = 146 for additive model, Nfs = 75 for recessive model, and Nfs = 49 for dominant model; for the ABCA1 M883I polymorphism, Nfs = 50 for allelic model), which reflected the minimum number of non-significant studies that would lead to the P-value to non-significant (P.0.05).

Discussion
To the best of our knowledge, this is the first comprehensive meta-analysis to date investigating the associations between the ABCA1 R219K and M883I polymorphisms and AS risk. Forty-two studies investigating the association between the ABCA1 R219K polymorphism and AS risk were combined , and 16 studies investigating the association between the ABCA1 M883I polymorphism and AS risk were combined  To make a more comprehensive analysis, subgroup analyses were performed based on ethnicity, atherosclerotic disease, source of controls and study type. For the ABCA1 R219K polymorphism, significant associations were found between this variant and the susceptibility to AS in the Asians group, CAD group, populationbased group, hospital-based group and the subgroup of casecontrol study, respectively. These results further strengthened the conclusion that the ABCA1 R219K K allele was a protective factor for AS. However, we did not find significant association between this variant and AS risk in Caucasians group, IS group and the subgroup of cohort study. For the ABCA1 M883I polymorphism, significant associations were found between this variant and the susceptibility to AS in the Asians group, CAD group, IS group, population-based group and the subgroup of case-control study, respectively. Similarly, no significant association was found between the ABCA1 M883I polymorphism and AS risk in Caucasians group and the subgroup of cohort study. Moreover, we also did not find significant association between this variant and AS risk in hospital-based group. One potential explanation is that controls in the hospital-based group may not truly represent the underlying source populations because the ABCA1 M883I polymorphism has been reported to be associated with various diseases. For instance, in 2004, Katzov et al. found that the ABCA1 polymorphism were involved in the process of cholesterol metabolism of the brain and may affect the risk of Alzheimer's disease [62]. In 2007, Kitjaroentham et al. indicated that the presence of ABCA1 polymorphism could be a risk factor for overweight/obesity among Thai males [63]. However, there was a noteworthy result in the IS group which suggested the ABCA1 M883I polymorphism was associated with an increased AS risk. This finding was inconsistent with the overall results, which may affect the interpretation of the final results. Throughout the IS group, it should be noted that the sample size in this group was really small (334 IS cases and 205 controls), which could increase the probability of false positives or false negatives. Therefore, this result should be interpreted with caution.
In the subgroup analysis based on ethnicity, we only obtained the positive result in the Asians. For Caucasians, we did not find significant association between the ABCA1 R219K and M883I polymorphisms and AS risk. In fact, the minor allele frequencies (MAF) of ABCA1 R219K and M883I are low among utah residents with Northern and Western European ancestry (CEU) (MAF = 0.210 and 0.133 in HapMap CEU) [64]. However, the ABCA1 R219K and M883I polymorphisms are common among Han Chinese in Beijing, China (CHB) and Japanese in Tokyo, Japan (JPT) (MAF = 0.453/0.434 and 0.255/0.438, respectively) [64]. Compared with the overall results, the inconsistent results of Caucasians group may be partly due to genetic diversity among ethnicities. Moreover, as AS is a multifactorial disease, in addition to genetic factors, environmental factors also play an important role in AS etiology. Thus, this discrepancy may also be caused by varied geographic distribution, linked to climate, diet, lifestyle and economic status.
Considering the studies without HWE or with low NOS score may influence the overall results, subsequent sensitivity analyses restricted to the studies with HWE or high NOS score were performed. Although the corresponding pooled OR was materially altered in the dominant model (OR = 0.84, 95% CI = 0.71-0.98, P = 0.03) of the ABCA1 M883I polymorphism (Based on HWE), this result further strengthened the conclusion that the ABCA1 M883I polymorphism was a protective factor for AS. As for the ABCA1 R219K polymorphism, the corresponding pooled ORs were not materially altered in all comparisons. These results suggested that the studies without HWE should be considered as a factor influencing the overall results. Significant heterogeneity existed in the present meta-analysis, either for the ABCA1 R219K polymorphism or for the ABCA1 M883I polymorphism. Heterogeneity should not be ignored and should be carefully factored in the interpretation of the finally results [65]. For the ABCA1 R219K polymorphism, the heterogeneity can be explained by the subtype of atherosclerotic diseases (CAD and IS) and study type (case-control and cohort study). For the ABCA1 M883I polymorphism, we did not indentify the sources of heterogeneity by the meta-regression. Common reasons of heterogeneity may include diversity in design, measurement errors, difference of ethnicity, or the interaction with other risk factors. Due to lack of original data, we did not perform further meta-regression analysis by the other factors.
For better interpreting the results, some limitations of this metaanalysis should be acknowledged. Firstly, there was significant heterogeneity in this meta-analysis. Heterogeneity may affect the precision of overall results, despite the use of appropriate metaanalytic techniques with random-effects model. Secondly, in the subgroup analyses, the sample sizes in some subgroup, such as the IS group and the hospital-based group of the ABCA1 M883I polymorphism, were relatively small, not having enough statistical power to explore the real association. Thirdly, publication bias was a potential problem that may bias the present results. Due to language restriction, some inevitable publication bias may exist in   the present meta-analysis. Moreover, AS is a complex disease, involving potential interactions among gene-gene and geneenvironment. A single gene polymorphism is insufficient to provide the complete explanation of genetic risk for AS. However, many eligible studies included in this meta-analysis did not consider the environmental factors.
In conclusion, the present meta-analysis suggested that the ABCA1 R219K and M883I polymorphisms were associated with the susceptibility to AS, especially in Asians. However, the result should be interpreted with caution because of its limitations. Further studies with large sample size, especially with the consideration of gene-gene and gene-environment interactions, will be needed to confirm our findings.