Gene 545 (2014) 175–183

Contents lists available at ScienceDirect

Gene

j ourna l homepage: www.e lsev ie r .com/ locate /gene

Association between the −786TNC 1polymorphism in the promoterregion of endothelial nitric oxide synthase (eNOS) and risk of coronaryartery disease: A systematic review and meta-analysis

Dan Liu a,1, Zhouqin Jiang b,1, Limeng Dai c, Xiaolin Zhang d, Chenghui Yan d, Yaling Han d,⁎a Graduate School of Third Military Medical University, Chongqing 400038, Chinab Department of Cardiology, Southwest Hospital, Third Military Medical University, Chongqing 400038, Chinac Department of Medical Genetics, College of Basic Medical Sciences, Third Military Medical University, Chongqing 400038, Chinad Cardiovascular Research Institute and Department of Cardiology, Shenyang Northern Hospital, Shenyang 110840, China

Abbreviations: BMI, body mass index; CAD, coronaryinterval; eNOS, endothelial nitric oxide synthase; HWE, Haodds ratio; PCR-RFLP, polymerase chain reaction-restriction⁎ Corresponding author at: Cardiovascular Research

Cardiology, Shenyang Northern Hospital, 83 Wenhua RTel./fax: +86 24 2391 1006.

E-mail address: hanyaling01@sina.com (Y. Han).1 The first two authors contributed equally to this work

0378-1119/$ – see front matter © 2013 Elsevier B.V. All rhttp://dx.doi.org/10.1016/j.gene.2013.09.099

a b s t r a c t

a r t i c l e i n f o

Article history:

Accepted 26 September 2013Available online 14 October 2013

Keywords:Endothelial nitric oxide synthase (eNOS)Coronary artery disease (CAD)PolymorphismMeta-analysis

Background: A variety of studies have evaluated the association between the −786TNC polymorphism in thepromoter region of endothelial nitric oxide synthase (eNOS) and risk of coronary artery disease (CAD). However,the results remain conflicting. To better understand the role of eNOS −786TNC polymorphism in CAD risk,we conducted a comprehensive systematic review and meta-analysis.Methods: Case–control, cohort or cross-sectional studies evaluating the association between eNOS −786TNCpolymorphism and CAD risk were searched in electronic databases of PubMed, ISI Web of Knowledge, Medline,Embase and Google Scholar Search (up to January 2013). Overall and subgroup analyses were performed. Oddsratio (OR) and 95% confidence interval (CI) were used to evaluate the association between eNOS −786TNC

polymorphism and CAD risk. Statistical analysis was performed with Review Manager 5.0 and STATA12.0.Results: Twenty-four studies were analyzed between 6192 CAD cases and 9281 healthy controls. The combinedresults of overall analysis showed significant positive associations between CAD risk and eNOS −786TNCpolymorphism in dominant model (OR = 1.45, 95% CI = 1.27–1.65), recessive model (OR = 1.37, 95% CI =1.20–1.56), homozygote comparison (OR = 1.64, 95% CI = 1.31–2.04), heterozygote comparison (TC vs. TT,OR = 1.39, 95% CI = 1.23–1.57; CC vs. TC, OR = 1.20, 95% CI = 1.04–1.37) and allele comparison (OR= 1.35,95% CI=1.21–1.50). On subgroup analysis based on the ethnicity of population (Caucasians, Asians and others),significant differences were found in all genetic models for Caucasians, similar associations existed in Asiansexcept heterozygote comparison (CC vs. TC). However, the associations were only found in dominant model,heterozygote comparison (TC vs. TT) and allele comparison for the populations named others.Conclusions: Our investigations demonstrate the significant associations between eNOS−786CNT polymorphismand CAD risk, and this polymorphism might become an early marker for the risk evaluation of CAD.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Coronary artery disease (CAD) is one of themajor causes ofmortality,accounting for approximately 30% of deaths worldwide (Cam et al.,2005; Kim et al., 2007). The genetic and environmental factorsplay important roles in the development of CAD (Salimi et al., 2006).

artery disease; CI, confidencerdy–Weinberg equilibrium; OR,fragment length polymorphism.Institute and Department ofoad, Shenyang 110840, China.

.

ights reserved.

Many epidemiological studies show that smoking, obesity, diabetesand hypertension are risk factors for CAD. However, some CADindividuals are not associated with the above risk factors, suggestingthat genetic factors could contribute to a predisposition to CAD.

Nitric oxide (NO), as an important endothelium-derived factor, issynthesized in the endothelial cells from L-arginine by the action of nitricoxide synthase (NOS) including inducible NOS, constitutive neuronalNOS and constitutive endothelial NOS (eNOS) (Gardemann et al., 2002;Palmer, 1993). NO participate in inhibiting the adhesion and aggregationof platelet,migration and growth of vascular smoothmuscle cell. LowNOrelease can cause several cardiovascular diseases such as atherosclerosis,hypertension and thrombosis, while high circulating NO concentrationis generally toxic.

The eNOS gene locating on chromosome 7q35–36 contains 26 exonsand encodes a 135kD protein including 1203 amino acids. The eNOSgene plays an important role in physiological and pathological process

Fig. 1. Flow diagram of study identification.

176 D. Liu et al. / Gene 545 (2014) 175–183

of cardiovascular system. Because NO availability is regulated at the levelof synthesis, the gene encoding eNOS is a candidate for cardiovasculardisease (Moncada and Higgs, 1993). The eNOS dysfunctions accelerateatherosclerosis in animal models and humans with atherosclerosis(Cayatte et al., 1994; Hingorani, 2001).

The−786TNCpolymorphism (rs2070744) is one of several promoterSNPs of eNOS that have been identifiedwith CAD risk, but the data aboutthis polymorphism from many individual studies are conflicting. Hanshowed that−786TNC polymorphism in the eNOS gene was associatedwith an increased risk of CAD in a Chinese population (Han et al., 2010),but Ragia found no association with CAD risk (Ragia et al., 2010).To better clarify the role of eNOS −786TNC polymorphism in CAD risk,we conducted a comprehensive systematic review and meta-analysisby collecting and analyzing the published data.

2. Materials and methods

2.1. Search strategy

Electronic databases of PubMed, ISI Web of Knowledge, Medline,Embase and Google Scholar Search were used to identify all publishedcase–control, cohort and cross-sectional studies evaluating theassociation between eNOS −786TNC polymorphism and CAD risk(up to January 2013). The Medical Subject Headings and key wordsused for search were “endothelial nitric oxide synthase,” “eNOS”,“NOS3,” “atherosclerotic heart disease,” “coronary artery disease,” and“myocardial infarction” combined with “genetic,” “polymorphism,”“mutation,” or “variation.” The references of all identified publicationswere hand-searched for additional studies. Authors were contacteddirectly regarding crucial data not reported in original articles. Abstracts,unpublished reports and articles written in non-English languages werenot included.

2.2. Inclusion and exclusion criteria

The inclusion criteriawere as follows: (1) independent case–control,cohort or cross-sectional design evaluating the association betweeneNOS −786TNC polymorphism and CAD risk was included, (2) thegenotype number or frequency was given in detail, (3) CAD patientdiagnosis was based on coronary angiography and healthy controlswere not CAD patients, (4) the genotype distribution of controlpopulation was in Hardy–Weinberg equilibrium (HWE).

The exclusion criteria were as follows: (1) studies with insufficientinformation were excluded, for example, genotype frequency ornumber not reported, or diagnosis of CAD not confirmed, (2) genotypedistribution of the control population didn't conform to HWE, (3) ifthe same population was included in previous studies, only the mostrecent or complete study was included after careful examination.

To minimize the bias and improve the reliability, two researchers(Dan Liu and Limeng Dai) searched the literatures and extracted datawith the inclusion and exclusion criteria independently and reacheda consensus.

2.3. Data extraction

For each study, information was extracted including the first author,publication year, country, ethnicity of study population, genotypenumber in cases/controls and genotypingmethod. For studies includingsubjects of different ethnicities, data were extracted separately andcategorized as Asians, Caucasians and others.

2.4. Statistical analysis

Association between eNOS −786TNC polymorphism and CAD riskwas evaluated by odds ratio (OR) and 95% confidence interval (CI). Inaddition to overall comparison, we also performed stratification analysis

based on ethnicity of study population and for myocardial infarction. Sixdifferent genetic models were used in our analysis: dominant model(CC+TC versus TT), recessive model (CC versus TC+TT), homozygotecomparison (CC versus TT), heterozygote comparison (TC versus TT,CC versus TC) and allele comparison (C versus T). The significance ofpooled ORs was tested by Z test (p b 0.05 was considered significant).Heterogeneity between studies was assessed using p value of Q test andI2 value. I2 was a value that could describe the percentage of variationacross studies, where 0–25% indicated no observed heterogeneity (thefixed effects model) and larger values showed increasing heterogeneity,with 25–50% regarded as low (the appropriate effect model accordingto p value of Q test and I2 value, the random effects model was used inour meta-analysis), 50–75% as moderate (the random effects model),and 75–100% as high (the random effects model). p N 0.05 for Q testindicated a lack of heterogeneity across studies, allowing to use thefixed effects model (Mantel and Haenszel, 1959), otherwise, the randomeffectsmodelwasused (DerSimonian and Laird, 1986). Theheterogeneitywas adjusted by subgroup analysis, sensitivity analysis and meta-regression. The funnel plot, Begg's test and harbord's modified test wereused to examine the publication bias (Egger et al., 1997). All p valueswere two-sided and all statistical analyses were performed using ReviewManager 5.0 and STATA12.0 software.

To ensure the reliability and accuracy of the results, two researchersentered the data into the software program independently and reacheda consensus.

3. Results

3.1. Study characteristics

By the inclusion and exclusion criteria, 392 articles were found, butonly 62 studies were preliminarily identified for further evaluation.After carefully evaluating the quality of the remaining 62 articles,we excluded 31 studies, of which 28 studies had unrelated data and 3studies were meta-analysis. After extracting data and reading the

Table 1Characteristics of literatures included in the meta-analysis.

Year Author Country Ethnicity Case Control HWE (p) in control HWE (p) in case Diagnose Genotyping method Source of control Average age(case/control)

TT TC CC N TT TC CC N

2012 Salimi et al. (2012) Iran Caucasian 111 102 28 241 168 82 11 261 0.8 0.53 CAD PCR-RFLP Age, gender-matched 53.32/51.82012 da Costa Escobar Piccoli et al. (2012) Brazil Other 53 63 16 132 50 52 12 114 0.78 0.68 ACS PCR-RFLP Community-based 61.3/61.82010 Ragia et al. (2010) Greece Caucasian 43 85 26 154 39 85 31 155 0.21 0.14 CAD PCR-RFLP Hospital-based 68/722010 Jaramillo et al. (2010) Chile Other 65 40 7 112 64 41 4 109 0.41 0.8 CAD PCR-RFLP Population-based 62/422010 Han et al. (2010) China Asian 185 110 17 312 226 76 8 310 0.6 0.9 CAD PCR-RFLP Population-based 61.96/60.542010 Bae et al. (2010) Korean Asian 111 27 0 138 139 22 0 161 0.35 0.2 CAD PCR-RFLP Age, gender-matched 60.86/60.542009 Meluzin et al. (2009) Czech Caucasian 132 186 44 362 25 28 4 57 0.3 0.07 CAD PCR-RFLP Hospital-based 62/542009 Alp et al. (2009) Turkey Other 70 68 8 146 66 42 14 122 0.08 0.09 CAD PCR-RFLP Age, gender-matched 59.3/57.32008 Ciftci et al. (2008) Turkey Other 16 8 6 30 21 8 2 31 0.33 0.02 CAD PCR-RFLP Hospital-based 60.45/59.002007 Rios et al. (2007) Brazil Other 69 61 17 147 76 37 8 121 0.24 0.53 CAD PCR-RFLP Hospital-based 55.6/52.52007 Kim et al. (2007) Korean Asian 110 35 2 147 182 38 0 220 0.16 0.67 CAD PCR-RFLP Hospital-based 56.5/52.72006 Tangurek et al. (2006) Turkey Other 63 73 23 159 31 15 6 52 0.07 0.8 CAD PCR-RFLP Hospital-based 59.6/56.82006 Dosenko et al. (2006) Ukraine Caucasian 105 80 36 221 40 38 5 83 0.3 0.003 ACS PCR-RFLP Age, gender-matched 58.5/54.22005 Rios et al. (2005) Brazil Caucasian 61 122 36 219 63 51 15 129 0.39 0.06 CAD PCR-RFLP Hospital-based 61.2/54.02004 Fatini et al. (2004) Italy Caucasian 123 247 107 477 199 260 78 537 0.64 0.42 ACS PCR-RFLP Population-based 68/662004 Agema et al. (2004) Dutch Caucasian 281 317 111 709 164 166 57 387 0.16 0.17 CAD PCR-RFLP Population-based 56.04/51.282003 Colombo et al. (2003) Italy Caucasian 54 124 58 236 47 71 20 138 0.41 0.43 CAD PCR-RFLP Hospital-based 60.8/53.52001 Granath et al. (2001) Australia Caucasian 220 267 86 573 245 285 93 623 0.5 0.73 CAD PCR-RFLP Community-based 43.9/40.92001 Alvarez et al. (2001) Spain Caucasian 52 81 37 170 120 137 43 300 0.7 0.61 CAD PCR-RFLP Population-based 42/392013 Arun Kumar et al. (2013) India Other 164 112 11 287 202 105 14 321 0.94 0.13 AMI PCR-RFLP Population-based 51.4/44.52009 Gluba et al. (2009) Poland Caucasian 177 86 15 278 83 44 7 134 0.71 0.29 AMI PCR-RFLP Healthy volunteer 41/542006 Jo et al. (2006) Korean Asian 91 36 2 129 643 150 10 803 0.71 0.46 AMI PCR-RFLP Population-based 59.08/62.042001 Takagi et al. (2001) Japan Asian 353 95 6 454 3127 748 43 3918 0.82 0.89 AMI Taqman Population-based Not mentioned2000 Nakayama et al. (2000) Japan Asian 276 78 5 359 179 16 0 195 0.55 0.85 AMI Allele-specific Hospital-based 62/61

177D.Liu

etal./Gene

545(2014)

175–183

Fig. 2. Forest plots for association between eNOS −786 polymorphism and CAD risk under dominant model (A: CC+TC versus TT) and recessive model (B: CC versus TC+TT).Random effect model was used. The squares and horizontal line represent the study-specific OR and 95% CI. The diamond represents the pooled results of OR and 95% CI.

178 D. Liu et al. / Gene 545 (2014) 175–183

remaining 31 studies (including 33 population data), 9 studies wereexcluded, with 6 studies without sufficient data and 3 studies deviatedfrom Hard–Weinberg equilibrium of control population. Finally, 24studies including 6192 CAD cases and 9281 controls were includedin our meta-analysis (Flow diagram shown in Fig. 1).

Information including the first author, publication year, country,ethnicity of study population, genotype number in cases/controls andgenotyping method was listed in Table 1. Different genotyping methodswere used in these studies including the classical polymerase chainreaction-restriction fragment length polymorphism (PCR-RFLP) (Agemaet al., 2004; Alp et al., 2009; Alvarez et al., 2001; Arun Kumar et al.,2013; Bae et al., 2010; Ciftci et al., 2008; Colombo et al., 2003; da CostaEscobar Piccoli et al., 2012; Dosenko et al., 2006; Fatini et al., 2004;Gluba et al., 2009; Granath et al., 2001; Han et al., 2010; Jaramillo et al.,2010; Jo et al., 2006; Kim et al., 2007; Meluzin et al., 2009; Ragia et al.,2010; Rios et al., 2005, 2007; Salimi et al., 2012; Tangurek et al., 2006),taqMan assay (Takagi et al., 2001), and allele-specific oligonucleotidemethod (Nakayama et al., 2000).

3.2. Meta-analysis results

The combined results of overall analysis showed significant positiveassociations between CAD risk and eNOS −786TNC polymorphism indominant model (CC+TC vs. TT, OR = 1.45, 95% CI = 1.27–1.65),recessive model (CC vs. TC+TT, OR = 1.37, 95% CI = 1.20–1.56),homozygote comparison (CC vs. TT, OR = 1.64, 95% CI = 1.31–2.04),heterozygote comparison (TC vs. TT, OR = 1.39, 95% CI = 1.23–1.57;CC vs. TC, OR=1.20, 95% CI=1.04–1.37) and allele comparison (C vs.T, OR=1.35, 95% CI=1.21–1.50). Forest plots on the basis of all studiesare shown in Figs. 2, 3 and 4.

In subgroup analysis based on the ethnicity of study population,significant associations were found in all genetic models for Caucasians

(dominant model, OR = 1.37, 95% CI = 1.12–1.67; recessive model,OR = 1.44, 95% CI = 1.14–1.82; homozygote comparison, OR = 1.71,95% CI = 1.26–2.31; heterozygote comparison (TC vs. TT, OR = 1.28,95% CI= 1.07–1.54; CC vs. TC, OR= 1.25, 95% CI= 1.07–1.46); allelecomparison, OR = 1.29, 95% CI = 1.11–1.50). However, positiveassociations for Asians were found in all genetic model expect forheterozygote comparison (CC vs. TC, OR = 1.32, 95% CI = 0.77–2.27),significant association for others were found in dominant model (OR=1.38, 95% CI = 1.15–1.67), heterozygote comparison (TC vs. TT, OR =1.39, 95% CI = 1.15–1.70) and allele comparison (OR= 1.26, 95% CI =1.08–1.46) (Table 2).

Because there were five studies involving acute myocardial infarction(AMI), a subgroup analysis for AMI was performed (Table 2). There weresignificant associations under dominant model (OR = 1.43, 95% CI =1.02–1.99), heterozygote comparison (TC vs. TT, OR = 1.41, 95% CI =1.02–1.97), and allele comparison (OR=1.36, 95% CI=1.01–1.83).

3.3. Heterogeneity analysis

Significant heterogeneity was found in overall comparisons andsubgroup analysis under dominant model, heterozygote comparison(TC vs. TT), homozygote comparison and allele comparison, we per-formed subgroup analysis, sensitivity analysis and meta-regressionto indicate the source of heterogeneity. According to the results ofmeta-analysis, several studies may be the source of the heterogeneity.Sensitivity analysis was performed to explain the heterogeneity. TheI2 decreased obviously and p value exceeded 0.05 after excluding thestudy of Salimi et al. (2012), Rios et al. (2005) and Nakayama et al.(2000) under dominant model (OR = 1.30, 95% CI = 1.20–1.42, I2 =35%, p = 0.06) and heterozygote comparison (OR = 1.27, 95% CI =1.16–1.38, TC vs. TT, I2 = 22%, p = 0.17). The I2 decreased obviouslyand p value exceeded 0.05 after excluding the study of Salimi et al.

Fig. 3. Forest plots for association between eNOS−786 polymorphism and CAD risk under heterozygote comparison (A: TC versus TT, random effect model, B: CC versus TC, fixed effectmodel). The squares and horizontal line represent the study-specific OR and 95% CI. The diamond represents the pooled results of OR and 95% CI.

179D. Liu et al. / Gene 545 (2014) 175–183

(2012), Rios et al. (2005), Fatini et al. (2004) and Nakayama et al.(2000) under homozygote comparison (OR = 1.35, 95% CI = 1.15–1.59, I2 = 30%, p= 0.10) and allele comparison (OR= 1.19, 95% CI =1.11–1.27, I2 = 44%, p = 0.09). The significance of pooled ORsand 95% CI in both overall comparison and subgroup analysis underthe dominantmodel, heterozygote comparison (TC vs. TT), homozygotecomparison and allele comparison was not influenced by omittingthe studies. Above results indicated that these studies were the majorsource of heterogeneity.

In addition, we also performed a meta-regression for explaining theheterogeneity. The results showed that source of control for each includedstudy independently contributed to the heterogeneity observed underdominant model (p = 0.035 b 0.05), heterozygote comparison (TC vs.TT, p=0.042b0.05) and allele comparison (p=0.034b0.05). However,genotyping methods, sample size and ethnicity of population were notstatistically associated with heterogeneity.

3.4. Publication bias analysis

Publication bias was assessed by performing funnel plot, harbord'smodified test and Begg's test under different genetic models. Theshape of the funnel plot appeared to be approximately symmetrical indominant and recessive models, the results of harbord's modified testand Begg's test suggested no evidence for publication bias [(dominantmodel: harbord's test p = 0.143, Begg's test p = 0.286), (recessivemodel: harbord's test p=0.190, Begg's test p=0.291)] (Fig. 5). Therewas no obvious publication bias in homozygote comparison (harbord's

test p=0.217, Begg's test p=0.597), heterozygote comparison [(TC vs.TT: harbord's test p=0.218, Begg's test p=0.535), (CC vs. TC: harbord'stest p=0.430, Begg's test p=0.224)] and allele comparison (harbord'stest p=0.079, Begg's test p=0.143).

4. Discussion

CAD is a complex disease influenced by genetic and environmentalfactors. Therefore, in addition to the known risk factors, genetic factorsmay play important roles in the pathogenesis and development of CAD(Wang et al., 1996). Currently, increasing evidences have showed thatgene polymorphisms are associated with CAD risk.

The eNOS can produce NO from L-arginine in the endothelial cell. Dueto the important roles of NOduring CAD and thephenomenon that eNOSdysfunctions were observed in human cardiovascular disease (Wilcoxet al., 1997), the eNOS gene has been identified as a further susceptibilityof CAD. The eNOS gene is encoded by 26 exons and includes severalSNP, some of which have functional relevance, the most relevanteNOS polymorphisms are the G894T variant, −786TNC in the promoterregion and 27 bp tandem repeats in intron 4 (4b/a) (Jia et al., 2007).Polymorphism −786TNC located in the promoter region of eNOSmay influence eNOS expression levels, lower eNOS mRNA and serumnitrite/nitrate levels have been found in individuals with the −786Callele (Miyamoto et al., 2000). To our knowledge, a variety of studiesconcerned the association between eNOS −786TNC polymorphism andsusceptibility to CAD, however, the results were inconsistent, so we

Fig. 4. Forest plots for association between eNOS −786 polymorphism and CAD risk under homozygote comparison (CC versus TT) and allele comparison (C versus T). Random effectmodel was used. The squares and horizontal line represent the study-specific OR and 95% CI. The diamond represents the pooled results of OR and 95% CI.

180 D. Liu et al. / Gene 545 (2014) 175–183

performed a meta-analysis on the association between eNOS −786TNCpolymorphism and CAD risk.

In our meta-analysis, overall analysis results showed that significantassociations between eNOS −786TNC polymorphism and CAD riskwere found in all six genetic models, identifying that the C allele ofeNOS −786TNC polymorphism had a statistically significant increasedCAD risk. Similarly, subgroup analysis based on the ethnicity of studypopulation showed that significant associations were found in all geneticmodels for Caucasians, positive associations for Asians were found in allgenetic model except for heterozygote comparison (CC vs. TC). However,significant associations for other populations were found in dominantmodel, heterozygote comparison (TC vs. TT) and allele comparison.Results for different countries and ethnicities were inconsistent, whichmight be caused by several factors such as environmental factors anddifferent genetic backgrounds. In addition, sample size in each includedstudy might influence the overall and subgroup analysis.

Notably, our meta-analysis results were different from a previousmeta-analysis, which revealed that no significant association betweeneNOS −786TNC polymorphism and risk of ischemic heart disease risk(OR = 1.06, 95% CI = 0.89–1.25) (Casas et al., 2004). Several reasonsmay explain for the difference. Firstly, the inclusion and exclusioncriteria were different. We included the studies with the strict criteriaas above described, such as three studies deviated fromHWE of controlpopulation were excluded (Alkharfy et al., 2010; Ameno et al., 2006;Rios et al., 2007), however, HWE of control in Casas's study was notmentioned. Secondly, the sample size in our study (6192 cases/9281controls) was different from Casas's study (2377 cases/7702 controls).Thirdly, we performed overall and subgroup analyses based on theethnicity of study population, however, only the overall analysis wasperformed in Casas's study.

The recent meta-analysis for eNOS−786TNC polymorphism showedan obviously significant association (OR = 0.79, 95% CI = 0.68–0.90)for T allele when the random effects model was adopted due to evidenceof significant heterogeneity (Li et al., 2010). Our results showed that Callele for this polymorphism could increase the risk of CAD. There wereseveral reasons for explaining the different results. Firstly, the inclusionand exclusion criteria were different, so the included study number inLi's article and ours was inconsistent. Secondly, the data in Li's studywas up to 2009 (14 studies), however, the data between 2010 and Jan2013 for this polymorphismwas supplemented in our study (24 studies).Thirdly, the sample size in our study (6192 cases/9281 controls) wasdifferent from the sample size in Li's study (3872 cases/3174 controls).Finally, the genetic models for analyzing the data in our study weredifferent from Li's study. All above contents could explain the differentresults between ours and Li's study.

Heterogeneity is an important problemwhen interpreting the resultsof our meta-analysis. Subgroup analysis, sensitivity analysis and meta-regression were used to explain the heterogeneity. The results of thesubgroup analysis revealed that the I2 for above geneticmodel decreasedobviously and p value exceeded 0.05 after excluding the studies (Fatiniet al., 2004; Nakayama et al., 2000; Rios et al., 2005; Salimi et al., 2012),indicating that these studies were the major source of heterogeneity.The significance of pooled ORs and 95% CI in both overall comparisonand subgroup analysis was not influenced. Meta-regression resultsshowed that source of control for each included study independentlycontributed to the heterogeneity observed under dominant model,heterozygote comparison and allele comparison. There were severalsources of control for each study including age and gender-matched,community-based, hospital-based and population-based, indicatingthat the subgroup analysis should be performed according to the source

Table 2Overall and subgroup analysis results of eNOS−786TNC polymorphism and CAD risk in six genetic models.

Category Na CC+TC versus TT CC versus TC+TT TC versus TT

OR (95% CI) I2 (%) p OR (95% CI) I2 (%) p OR (95% CI) I2 (%) p

Overall 24 (Agema et al., 2004; Alp et al., 2009; Alvarez et al.,2001; ArunKumar et al., 2013; Bae et al., 2010; Ciftci et al.,2008; Colombo et al., 2003; da Costa Escobar Piccoli et al.,2012; Dosenko et al., 2006; Fatini et al., 2004; Glubaet al., 2009; Granath et al., 2001; Han et al., 2010;Jaramillo et al., 2010; Jo et al., 2006; Kim et al.,2007; Meluzin et al., 2009; Nakayama et al.,2000; Ragia et al., 2010; Rios et al., 2005, 2007;Salimi et al., 2012; Takagi et al., 2001;Tangurek et al., 2006)

1.45 (1.27–1.65) 57 0.0003 1.37 (1.20–1.56)b 33 0.07 1.39 (1.23–1.57) 49 0.004

Caucasians 11 (Agema et al., 2004; Alvarez et al., 2001;Colombo et al., 2003; Dosenko et al., 2006;Fatini et al., 2004; Gluba et al., 2009;Granath et al., 2001; Meluzinet al., 2009; Ragia et al., 2010;Rios et al., 2005; Salimi et al., 2012)

1.37 (1.12–1.67) 68 0.0006 1.44 (1.14–1.82) 51 0.03 1.28 (1.07–1.54) 58 0.008

Asian 6 (Bae et al., 2010; Han et al., 2010; Jo et al., 2006;Kim et al., 2007; Nakayama et al., 2000;Takagi et al., 2001)

1.69 (1.26–2.27) 67 0.009 1.86 (1.11–3.09)b 0 0.59 1.64 (1.24–2.17) 62 0.02

Other 7 (Alp et al., 2009; Arun Kumar et al., 2013;Ciftci et al., 2008; da Costa EscobarPiccoli et al., 2012; Jaramillo et al.,2010; Rios et al., 2007;Tangurek et al., 2006)

1.38 (1.15–1.67)b 2 0.41 1.14 (0.80–1.62)b 25 0.24 1.39 (1.15–1.70)b 0 0.45

AMI 5 (Arun Kumar et al., 2013; Gluba et al., 2009;Jo et al., 2006; Nakayama et al., 2000;Takagi et al., 2001)

1.43 (1.02–1.99) 75 0.003 1.13 (0.71–1.79)b 0 0.78 1.41 (1.02–1.97) 73 0.006

Na CC+TC versus TT CC versus TC+TT TC versus TT

OR (95% CI) I2 (%) p OR (95% CI) I2 (%) p OR (95% CI) I2 (%) p

Overall 24 (Agema et al., 2004; Alp et al., 2009;Alvarez et al., 2001; Arun Kumar et al., 2013;Bae et al., 2010; Ciftci et al., 2008; Colombo et al., 2003;da Costa Escobar Piccoli et al., 2012; Dosenko et al., 2006;Fatini et al., 2004; Gluba et al., 2009; Granath et al., 2001;Han et al., 2010; Jaramillo et al., 2010; Jo et al., 2006;Kim et al., 2007; Meluzin et al., 2009; Nakayama et al.,2000; Ragia et al., 2010; Rios et al., 2005, 2007;Salimi et al., 2012; Takagi et al., 2001;Tangurek et al., 2006)

1.20 (1.04–1.37)b 9 0.34 1.64 (1.31–2.04) 47 0.007 1.35 (1.22–1.50) 62 b0.0001

Caucasians 11 (Agema et al., 2004; Alvarez et al., 2001;Colombo et al., 2003; Dosenko et al., 2006; Fatini et al.,2004; Gluba et al., 2009; Granath et al., 2001;Meluzin et al., 2009; Ragia et al., 2010;Rios et al., 2005; Salimi et al., 2012)

1.25 (1.07–1.46)b 22 0.24 1.71 (1.26–2.31) 65 0.001 1.29 (1.11–1.50) 71 0.0002

Asian 6 (Bae et al., 2010; Han et al., 2010; Jo et al., 2006;Kim et al., 2007; Nakayama et al., 2000;Takagi et al., 2001)

1.32 (0.77–2.27)b 0 0.82 1.92 (1.11–3.31)b 0 0.5 1.62 (1.23–2.13) 68 0.008

Other 7 (Alp et al., 2009; Arun Kumar et al., 2013;Ciftci et al., 2008; da Costa Escobar Piccoli et al., 2012;Jaramillo et al., 2010; Rios et al., 2007;Tangurek et al., 2006)

0.90 (0.62–1.31)b 20 0.28 1.34 (0.93–1.92)b 24 0.25 1.26 (1.08–1.46)b 21 0.27

AMI 5 (Arun Kumar et al., 2013; Gluba et al., 2009;Jo et al., 2006; Nakayama et al., 2000;Takagi et al., 2001)

0.96 (0.60–1.55)b 0 0.92 1.20 (0.75–1.90)b 0 0.75 1.36 (1.01–1.83) 76 0.002

a The number of study.b The Fixed-effects model was chosen to summarize the result.

181D. Liu et al. / Gene 545 (2014) 175–183

of the control. However, the number of study for each source ofcontrol was less, there was no necessary to perform subgroup analysis.More studies should be done in further studies to elucidate the effect ofcontrol source to the heterogeneity of meta-analysis.

Some limitations in our analysis should be acknowledged.Firstly, although we collected all the eligible studies, the sample size ofthe studies included was not large enough especially in subgroupanalysis, which could increase the probability of false positives or falsenegatives. Secondly, many factors participate in CAD progression, wedidn't analyze the data based on age and gender due to the limiteddata. Thirdly, gene–gene and gene–environment interactions were notanalyzed. Specific environmental and lifestyle factors may alter the

association. It is necessary to evaluate the roles of some factors such assmoking status, bodymass index (BMI), hypertension, diabetesmellitusin CAD risk. Finally, although the funnel plot, harbord's modified testand Begg's test did not show any publication bias, the influence of biasin the present analysis could not be completely excluded. For example,studies with positive results are more easily published than those withnegative results, and only studies published in English are included.

5. Conclusions

The results in our meta-analysis demonstrate that C allele of eNOS−786TNC polymorphism could increase CAD risk compared to T allele,

Fig. 5. Funnel plot for eNOS −786 polymorphism and CAD risk under dominant model (A: CC+TC versus TT) and recessive model (B: CC versus TC+TT).

182 D. Liu et al. / Gene 545 (2014) 175–183

although some results are limited by the small number of studies.The eNOS −786TNC polymorphism may become early marker for riskevaluation of CAD. Moreover, further studies with large sample size,especially in subgroup analysis, should be done to confirm our findings.

Funding

This work was supported by the key project of National NaturalScience Foundation of China grant (No. 81130072) and China Ministryof Science and Technology Project 973 (No. 2011CB512111).

Conflicts of interest

We declare that there are no conflicts of interest in this study.

References

Agema, W.R., et al., 2004. An integrated evaluation of endothelial constitutive nitric oxidesynthase polymorphisms and coronary artery disease in men. Clin. Sci. (Lond.) 107,255–261.

Alkharfy, K.M., Al-Daghri, N.M., Al-Attas, O.S., Alokail, M.S., Draz, H.M., Hussain, T., 2010.Endothelial nitric oxide synthase gene polymorphisms (894GNT and −786TNC) andrisk of coronary artery disease in a Saudi population. Arch. Med. Res. 41, 134–141.

Alp, E., et al., 2009. Lack of association between matrix metalloproteinase-9 andendothelial nitric oxide synthase gene polymorphisms and coronary artery diseasein Turkish population. DNA Cell Biol. 28, 343–350.

Alvarez, R., et al., 2001. Association between the NOS3 (−786T/C) and the ACE (I/D) DNAgenotypes and early coronary artery disease. Nitric Oxide 5, 343–348.

Ameno, K., et al., 2006. Autopsy and postmortem examination case study on genetic riskfactors for cardiac death: polymorphisms of endothelial nitric oxide synthase geneGlu298Asp variant and T-786C mutation, human paraoxonase 1 (PON1) gene andalpha2beta-adrenergic receptor gene. Vojnosanit. Pregl. 63, 357–361 (discussion62–3).

Arun Kumar, A.S., Umamaheswaran, G., Padmapriya, R., Balachandar, J., Adithan, C., 2013.Endothelial nitric oxide synthase gene polymorphisms and the risk of acutemyocardial infarction in a South Indian population. Mol. Biol. Rep. 40, 1275–1281.

Bae, J., et al., 2010. Association of endothelial nitric oxide synthase polymorphisms withcoronary artery disease in Korean individuals with or without diabetes mellitus.Exp. Ther. Med. 1, 719–724.

Cam, S.F., et al., 2005. The G894T polymorphism on endothelial nitric oxide synthase geneis associated with premature coronary artery disease in a Turkish population.Thromb. Res. 116, 287–292.

Casas, J.P., Bautista, L.E., Humphries, S.E., Hingorani, A.D., 2004. Endothelial nitric oxidesynthase genotype and ischemic heart disease: meta-analysis of 26 studies involving23028 subjects. Circulation 109, 1359–1365.

Cayatte, A.J., Palacino, J.J., Horten, K., Cohen, R.A., 1994. Chronic inhibition of nitricoxide production accelerates neointima formation and impairs endothelial functionin hypercholesterolemic rabbits. Arterioscler. Thromb. 14, 753–759.

Ciftci, C., et al., 2008. Association of endothelial nitric oxide synthase promoter region (T-786C) gene polymorphism with acute coronary syndrome and coronary heartdisease. Lipids Health Dis. 7, 5.

Colombo, M.G., et al., 2003. Endothelial nitric oxide synthase gene polymorphisms andrisk of coronary artery disease. Clin. Chem. 49, 389–395.

da Costa Escobar Piccoli, J., et al., 2012. Interaction between endothelial nitric oxidesynthase gene polymorphisms (−786TNC, 894GNT and intron 4 a/b) andcardiovascular risk factors in acute coronary syndromes. Arch. Med. Res. 43, 205–211.

DerSimonian, R., Laird, N., 1986. Meta-analysis in clinical trials. Control. Clin. Trials 7,177–188.

Dosenko, V.E., Zagoriy, V.Y., Lutay, Y.M., Parkhomenko, A.N., Moibenko, A.A., 2006. Allelicpolymorphism in the promoter (T→C), but not in exon 7 (G→T) or the variablenumber tandem repeat in intron 4, of the endothelial nitric oxide synthase gene ispositively associated with acute coronary syndrome in the Ukrainian population.Exp. Clin. Cardiol. 11, 11–13.

Egger, M., Davey Smith, G., Schneider, M., Minder, C., 1997. Bias in meta-analysis detectedby a simple, graphical test. BMJ 315, 629–634.

Fatini, C., et al., 2004. Influence of endothelial nitric oxide synthase gene polymorphisms(G894T, 4a4b, T-786C) and hyperhomocysteinemia on the predisposition to acutecoronary syndromes. Am. Heart J. 147, 516–521.

Gardemann, A., Lohre, J., Cayci, S., Katz, N., Tillmanns, H., Haberbosch, W., 2002. The Tallele of the missense Glu(298)Asp endothelial nitric oxide synthase gene poly-morphism is associated with coronary heart disease in younger individuals withhigh atherosclerotic risk profile. Atherosclerosis 160, 167–175.

Gluba, A., Banach, M., Rysz, J., Piotrowski, G., Fendler, W., Pietrucha, T., 2009.Is polymorphism within eNOS gene associated with the late onset of myocardialinfarction? A pilot study. Angiology 60, 588–595.

Granath, B., Taylor, R.R., van Bockxmeer, F.M., Mamotte, C.D., 2001. Lack of evidencefor association between endothelial nitric oxide synthase gene polymorphisms andcoronary artery disease in the Australian Caucasian population. J. Cardiovasc. Risk 8,235–241.

Han, Y., Xu, W., Zhang, W., Liu, N., Ji, Y., 2010. T-786C polymorphism in the endothelialnitric oxide synthase gene is associated with increased risk of coronary artery diseasein a Chinese population. Pharmacology 85, 211–216.

Hingorani, A.D., 2001. Polymorphisms in endothelial nitric oxide synthase andatherogenesis: John French Lecture 2000. Atherosclerosis 154, 521–527.

Jaramillo, P.C., Lanas, C., Lanas, F., Salazar, L.A., 2010. Polymorphisms of the NOS3 gene inSouthern Chilean subjects with coronary artery disease and controls. Clin. Chim. Acta411, 258–262.

Jia, C., Liu, T., Liu, Z., Li, M., Hu, M., 2007. Joint effects of eNOS gene T-786C and ADH2Arg47His polymorphisms on the risk of premature coronary artery disease. Thromb.Res. 120, 679–684.

Jo, I., et al., 2006. Interaction between −786TC polymorphism in the endothelial nitricoxide synthase gene and smoking for myocardial infarction in Korean population.Clin. Chim. Acta 365, 86–92.

Kim, I.J., et al., 2007. Influence of endothelial nitric oxide synthase gene polymorphisms(−786TNC, 4a4b, 894GNT) in Korean patients with coronary artery disease. Thromb.Res. 119, 579–585.

Li, J., Wu, X., Li, X., Feng, G., He, L., Shi, Y., 2010. The endothelial nitric oxide synthase gene isassociated with coronary artery disease: a meta-analysis. Cardiology 116 (4), 271–278.

Mantel, N., Haenszel, W., 1959. Statistical aspects of the analysis of data fromretrospective studies of disease. J. Natl. Cancer Inst. 22, 719–748.

Meluzin, J., Vasku, A., Kincl, V., Panovsky, R., Sramkova, T., 2009. Associationof coronary arterydisease, erectile dysfunction, and endothelial nitric oxide synthase polymorphisms.Heart Vessels 24, 157–163.

Miyamoto, Y., et al., 2000. Replication protein A1 reduces transcription of the endothelialnitric oxide synthase gene containing a −786T→C mutation associated withcoronary spastic angina. Hum. Mol. Genet. 9, 2629–2637.

Moncada, S., Higgs, A., 1993. The L-arginine-nitric oxide pathway. N. Engl. J. Med. 329,2002–2012.

Nakayama, M., et al., 2000. T(−786)→C mutation in the 5′-flanking region of theendothelial nitric oxide synthase gene is associated with myocardial infarction,especially without coronary organic stenosis. Am. J. Cardiol. 86, 628–634.

Palmer, R.M., 1993. The L-arginine: nitric oxide pathway. Curr. Opin. Nephrol. Hypertens.2, 122–128.

Ragia, G., et al., 2010. Endothelial nitric oxide synthase gene polymorphisms −786TNCand 894GNT in coronary artery bypass graft surgery patients. Hum. Genomics 4,375–383.

Rios, D.L., Callegari-Jacques, S.M., Hutz, M.H., 2005. Endothelial nitric oxide synthaseand fractalkine chemokine receptor polymorphisms on angiographically assessedcoronary atherosclerosis. Clin. Chim. Acta 362, 138–146.

Rios, D.L., et al., 2007. Smoking-dependent and haplotype-specific effects of endothelial nitricoxide synthase gene polymorphisms on angiographically assessed coronary arterydisease in Caucasian- and African-Brazilians. Atherosclerosis 193, 135–141.

183D. Liu et al. / Gene 545 (2014) 175–183

Salimi, S., Firoozrai, M., Nourmohammadi, I., Shabani, M., Mohebbi, A., 2006. Endothelialnitric oxide synthase gene intron4 VNTR polymorphism in patients with coronaryartery disease in Iran. Indian J. Med. Res. 124, 683–688.

Salimi, S., et al., 2012. Association of plasma nitric oxide concentration and endothelialnitric oxide synthase T-786C gene polymorphism in coronary artery disease.Pathophysiology 19, 157–162.

Takagi, S., Goto, Y., Nonogi, H., Baba, S., Iwai, N., 2001. Genetic polymorphismsof angiotensin converting enzyme (I/D) and endothelial nitric oxide synthase(T(−788)C) genes in Japanese patients with myocardial infarction. Thromb.Haemost. 86, 1339–1340.

Tangurek, B., et al., 2006. The relationship between endothelial nitric oxide synthase genepolymorphism (T-786 C) and coronary artery disease in the Turkish population.Heart Vessels 21, 285–290.

Wang, X.L., Sim, A.S., Badenhop, R.F., McCredie, R.M., Wilcken, D.E., 1996.A smoking-dependent risk of coronary artery disease associated witha polymorphism of the endothelial nitric oxide synthase gene. Nat. Med. 2,41–45.

Wilcox, J.N., et al., 1997. Expression of multiple isoforms of nitric oxide synthasein normal and atherosclerotic vessels. Arterioscler. Thromb. Vasc. Biol. 17,2479–2488.