- Email: [email protected]
The Connexin37 Gene C1019T Polymorphism and Risk of Coronary Artery Disease: A Meta-analysis
Archives of Medical Research 45 (2014) 21e30
ORIGINAL ARTICLE
The Connexin37 Gene C1019T Polymorphism and Risk of Coronary Artery Disease: A Meta-analysis Zhijun Wu,a Yuqing Lou,b Wei Jin,a Yan Liu,a Lin Lu,a Qiujing Chen,a and Ruiyan Zhanga a
Department of Cardiology, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, PR China Department of Pulmonary Diseases, Shanghai Chest Hospital, Shanghai Jiao Tong University, Shanghai, PR China
b
Received for publication July 3, 2013; accepted October 18, 2013 (ARCMED-D-13-00355).
Background and Aims. Mounting data have emerged suggesting that the Connexin37 C1019T polymorphism increases susceptibility to coronary artery disease (CAD). However, previous studies yielded conflicting results. In the current study, a comprehensive meta-analysis was performed to investigate whether the C1019T polymorphism is associated with CAD risk. Methods. A total of 11 studies examining the C1019T polymorphism and CAD were identified using MEDLINE, Embase, CNKI, Wanfang and CBM, in which 5535 CAD patients and 5626 controls were analyzed. A random-effects model was used to calculate odd ratios and confidence intervals, while addressing between-study heterogeneity. Publication bias was weighed using the Egger’s test, Begg-Mazemdar test and funnel plot. Results. In genetic models with striking heterogeneity, the risk of CAD was not associated with the C1019T polymorphism (allele comparison: p 5 0.34, OR 5 1.11, 95% CI 0.90e1.36). Stratification by disease endpoints indicated that the 1019T allele was significantly associated with myocardial infarction (MI) (allele comparison: p !0.001, OR 5 1.59, 95% CI 1.24e2.03). Further meta-regression analysis indicated that a large proportion of heterogeneity was probably due to the varying proportions of diabetes mellitus (DM) across studies ( p 5 0.014). Conclusions. Our results indicated that the C1019T polymorphism may be a moderate risk factor for MI and that DM was likely a potential source of between-study heterogeneity. Ó 2014 IMSS. Published by Elsevier Inc. Key Words: Connexin, Coronary artery disease, Myocardial infarction, Polymorphism Metaanalysis.
Introduction A considerable portion of coronary artery disease (CAD) and myocardial infarction (MI) events cannot be characterized by the traditional risk factors. Diseases and conditions such as hypertension (HTN), diabetes (DM) and hyperlipidemia, or other characteristics and habits such as being overweight and smoking are generally regarded as contributors to cardiovascular diseases (1). In addition, a broad-spectrum of Address reprint requests to: Ruiyan Zhang, MD, PhD, Department of Cardiology, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, No 197, Ruijin Er Road, Shanghai, 200025, PR China; Phone: (þ86) 21-64370045; FAX: (þ86) 21-64457177; E-mail: zhangruiyan@ 263.net
underlying pathological processes, ranging from lipoprotein deposits in the arterial intima to blood thrombosis formation due to plaque rupture, have been associated with cardiovascular disease (2). Among the major pathways, endothelial integrity and persistent inflammation appear to be key factors that play roles in the biological etiology of CAD and MI (3). The endothelial barrier is supported by gap- junctional proteins and prevents untoward diapedesis and penetration of low-density lipoprotein cholesterol (4). Connexins are members of a family of endothelial gapjunctional proteins and are expressed in the vascular wall. Six connexin subunits form a hemi-channel called a connexon in the plasma membrane. Connexons dock with their counterparts in the plasma membrane of an adjacent cell
0188-4409/$ - see front matter. Copyright Ó 2014 IMSS. Published by Elsevier Inc. http://dx.doi.org/10.1016/j.arcmed.2013.12.001
22
Wu et al./ Archives of Medical Research 45 (2014) 21e30
and assemble an intercellular channel called a gap junction (5,6), Gap junctions contribute to the maintenance of cell growth and differentiation (7). In the human genome, the connexin gene family is comprised of 21 genes. Different connexins are named according to the molecular weight that is predicted by their respective cDNAs (8). Connexin 37 (Cx 37) has a unique expression in endothelial and macrophage cells and plays a key role in atherogenesis (9,10). Deletion of Cx37 accelerates atherosclerosis in Cx37/ApoE/mice compared with the Cx37þ/þApoE/ controls, suggesting that Cx37 may exert an anti-atherosclerosis effect on ApoE/ mice (11). Additionally, there is a higher proportion of Cx37deficient monocytes and macrophages among leukocytes that are related to atherosclerotic plaques (12). The human Cx37 gene is located at chromosome 1p35.1 (13). One of the most common polymorphisms of the Cx37 gene is a cytosine-to-thymine replacement at position 1019 resulting in a substitution of serine for proline at codon 319 (P319S) in the regulatory C-terminus of Cx37 protein. The Cx37 polymorphism C1019T has been identified as a prognostic marker for atherothrombosis (14,15). The C1019T polymorphism has been significantly associated with a 3-year survival after acute coronary syndrome (ACS) (15). Therefore, it is of added interest to identify whether the C1019T polymorphism is related to the risk of CAD or MI. However, published association studies have reported either inconclusive or conflicting results. Generally speaking, insufficient sample size, marked racial and ethnic differences, unadjusted environmental confounders and genetic background noise are likely at least partially responsible for the current discordance among results. Meta analysis is an ideal tool to combine paradoxical results from different studies (16) in order to estimate the major effect with enhanced precision. In the current study, a meta-analysis is done incorporating the available data to date, to investigate whether the Cx37 C1019T polymorphism is a risk factor for the CAD. Between-study heterogeneity and publication bias are also addressed.
Materials and Methods Search Strategy A literature search strategy was undertaken using PubMed/ MEDLINE, Embase, the China National Knowledge Infrastructure (CNKI), Wanfang and the Chinese Biological Medicine (CBM) electronic databases up to June 2013. Combinations of the following key words were used as medical subject heading (MeSH) terms and text words: ‘C1019T’ or ‘Pro319Ser’, ‘connexin 37’, ‘gap junction protein alpha-4’ or ‘GJA4’ and ‘coronary’ or ‘ischemic heart disease’ or ‘myocardial infarction’ or ‘atherosclerosis’. The full texts of potentially appropriate articles were scrutinized to decide whether the data would be examined further.
The bibliographies of relevant studies, reviews and previous meta-analyses were reviewed to identify additional studies that were not indexed by the initial searches. Hand searches and the related search functions in PubMed were also used. If the allele or genotype frequencies were incompletely reported, the corresponding authors we were contacted to obtain the raw data. Selection Criteria Studies related to the association of the Cx37 C1019T polymorphism with CAD risk were provisionally included if (1) they were published in English or Chinese journals or their supplements; (2) they provided sufficient information on the C1019T polymorphism and studies without controls were excluded; (3) the genotype frequency of the control population conformed to Hardy-Weinberg Equilibrium (HWE). There were no restrictions on ethnicity or country in which the studies were conducted. When an article involved more than one subpopulation, each population was considered as a separate study. When multiple studies were derived from the same population, only the largest scale study was included to avoid double calculation of the odd ratios (ORs). The diagnosis of MI was done on the basis of typical electrocardiographic changes (both ST-segment elevation and non-ST-segment elevation changes) and a positive troponin blood test (17). A diagnosis of acute coronary syndrome (ACS) was based on unstable angina pectoris, as well as acute and subacute MI (18). CAD was defined as a stenosis diameter of $50% in any of the major coronary arteries or in the left main trunk as documented by coronary angiography (19). Data Extraction Two authors (ZW and YL) independently obtained the full texts of the eligible articles based on the titles and abstracts. The necessary data was extracted from the selected studies and entered into separate databases in duplicate. Any disagreements were resolved at a consensus meeting. Data were collated according to different cohorts: first author’s name, publication year, ethnicity, geographical location, study design, population source, disease outcome, sample size and clinical characteristics (e.g., age, body mass index [BMI], circulating lipid profiles, the proportion of males, hypertension [HTN], diabetes mellitus [DM] and smoking status). The C1019T genotype frequency and its consistency with HWE were also collected for both patients and controls. Continuous variables are manifested as mean standard deviation (SD) or median (5th and 95th percentiles). Statistical Analysis Allele frequencies of the C1019T polymorphism were determined by the allele counting method. Data were combined
Connexin37 Gene C1019T Polymorphism and Risk of Coronary Artery Disease
23
Figure 1. Flow diagram of the search strategy and study selection for the meta-analysis.
and reported as weighted ORs corresponding to the 95% confidence intervals (CIs) of the association between the C1019T polymorphism and CAD risk. Four genetic models were used: allele comparison (T vs. C), the dominant genetic model (CTþTT vs. CC), the recessive genetic model (TT vs. CTþCC) and homozygote comparison (TT vs. CC). A random-effects model using DerSimonian & Laird method was used to estimate the individual effect size and to modulate study weights based on in-study variance. The method does not account for heterogeneity, thus the Mantel-Haenszel model was used to assess heterogeneity (20). In addition, Cochran’s chi-square based Q statistical test was used to evaluate within- and between-study heterogeneity in which p !0.10 was regarded as significant (21). The I2 statistic was used to evaluate the degree of inconsistency in the studies or, in other words, to estimate the percentage of between-study variability due to heterogeneity
rather than due to chance (22,23). The I2 ranged from 0e100% where a value of 0% indicated no observed heterogeneity and increasing values indicated increasing degrees of heterogeneity. Pooled OR significance was evaluated by the Z test and p !0.05 was considered to be significant. The pre-specified characteristics for evaluation of interstudy heterogeneity sources were as follows: ethnicity (Asian and Caucasian), endpoints (CAD, ACS and MI), study design (matched and not reported), and population source (hospital-based [H-B] and population-based [P-B]). Additionally, a meta-regression was done to evaluate the extent to which different covariates, including the characteristics mentioned above, explained the genetic heterogeneity among the individual ORs. Sensitivity analyses were performed to identify the most influential studies that could contribute to the possibility of biased effects estimation. Each study was successive
24
Table 1. Baseline characteristics of all qualified studies included in the meta-analysis
Ethnicity
Geographic location Source Endpoint
Year
Feng W
2010 Asian
China
H-B
CAD
Han YL
2007 Asian
China
H-B
CAD
Czekh
P-B
ACS
Hubacek JA 2010 Caucasian
Study design
Status
Not mentioned Cases Controls Not mentioned Cases Controls Not mentioned Cases Controls
Listi F
2005 Caucasian
Morgan TM 2007 Caucasian
Italy
P-B
MI
US
H-B
ACS
Not mentioned Cases Controls Matched Cases Controls
Seifi M
2013 Caucasian
Iran
P-B
MI
Xie XJ
2006 Asian
China
H-B
CAD
Yamada Y
2002 Asian
Japan
H-B
MI
Yeh HI
2001 Asian
China
H-B
CAD
Zhang CL
2009 Asian
China
H-B
MI
Zhang SR
2007 Asian
China
H-B
CAD
Not mentioned Cases Controls Not mentioned Cases Controls Not mentioned Cases Controls matched Cases Controls Not mentioned Cases Controls Not mentioned Cases Controls
Age, year 64.4 11.4 57.0 10.4 59.3 10.6 52.9 10.4 55.2 7.5 (M) 62.6 8.5 (F) 49.0 10.8 (M) 48.6 10.6 (F) 40.0 (20.0e46.0) 39.0 (20.0e55.0) 60.7 12.5 (M) 63.1 13.2 (F) 60.0 12.1 (M) 61.8 12.8 (F) 58.0 13.0 57.0 13.0 61.0 10.0 60.0 10.0 62.1 10.1 62.0 10.4 61.0 11.0 60.0 11.0 57.2 12.6 50.6 11.1 68.0 10.9 66.7 8.6
Gender, M (%) HTN, % DM, % Smoke, % 50.0 47.4 77.3 59.0 72.6
20.8 23.7 59.4 46.6 55.5
15.1 15.8 17.7 12.9 39.0
37.7 26.3 40.6 30.2 73.0
46.5
36.7
7.8
26.7
100.0 100.0 67.8
31.0 60.1
17.0 23.8
-a 33.1
60.6
51.2
11.9
13.2
64.0 64.3 81.3 61.5 100.0 100.0 67.8 68.6 72.8 60.5 78.6 58.1
41.3 37.6 47.0 57.3 80.5 55.7
34.7 16.2 23.0 21.0 46.1 17.5
42.7 21.4 57.8 55.2 49.0 47.0 49.7 20.6
BMI, kg/m2 24.2 24.2 25.4 25.0 28.5 28.8 28.2 27.6
29.1 29.9 27.9 27.7 26.3 26.7 25.0 25.0 23.6 23.6 25.2 25.5
24.5 23.5
1.7 1.5 3.0 3.2 4.3 5.6 4.0 5.5 5.5 6.9 5.0 6.9 4.7 4.9 3.0 4.0 2.9 2.7 3.2 3.7 3.7 5.1
TC, mmol/L
4.0 3.9 4.6 4.5 (M) 5.2 (F) 5.4 (M) 5.8 (F) 5.8
(M) (F) (M) (F) 5.1 4.9 5.0 4.6
5.2 5.0 4.8 4.4
0.9 0.8 1.0 0.8 1.2 1.3 1.1 1.2 1.1 1.0 1.1 0.9 1.1 1.2 1.2 1.4 -
Tg, mmol/L
1.3 1.3 2.0 2.0 (M) 2.1 (F) 1.9 (M) 2.0 (F) 1.5
1.9 1.8 1.6 1.5
2.0 1.9 1.6 1.5
0.4 0.5 1.4 1.6 1.5 1.2 1.3 0.8 1.1 1.1 1.2 1.3 1.3 1.3 0.7 0.9 -
(M) (F) (M) (F)
P-B, population-based study; H-B, hospital-based study; CAD, coronary artery disease; MI, myocardial infarction; ACS, acute coronary syndrome; M(%), male (percent); F, female; HTN, hypertension; DM, diabetes mellitus; BMI, body mass index; TC, total cholesterol; Tg, triglyceride. a Data not available; age and BMI are expressed as mean SD (standard deviation) or median (5th and 95th percentiles).
Wu et al./ Archives of Medical Research 45 (2014) 21e30
First author
Connexin37 Gene C1019T Polymorphism and Risk of Coronary Artery Disease
25
Table 2. Sample size, distribution of the C1019T allele frequencies and genotypes among CAD cases and controls, and p value of HWE in controls Sample size
T allele, %
C allele, %
TT genotype
CT genotype
CC genotype
HWE,
First author
Cases
Controls
Cases
Controls
Cases
Controls
Cases
Controls
Cases
Controls
Cases
Controls
p
Feng W Han YL Hubacek JA Listi F Morgan TM Seifi M Xie XJ Yamada Y Yeh HI Zhang CL Zhang SR Total
106 502 1399 97 811 200 150 1784 177 136 173 5535
38 410 2559 196 650 185 117 1074 102 147 148 5626
43.9 51.4 29.9 43.8 29.6 53.8 25.0 19.0 10.7 36.0 35.3 29.2
59.2 58.7 31.6 34.4 31.8 40.0 21.4 15.6 20.1 20.4 23.7 30.1
56.1 48.6 70.1 56.2 70.4 46.3 75.0 81.0 89.3 64.0 64.7 70.8
40.8 41.3 68.4 65.6 68.2 60.0 78.6 84.4 79.9 79.6 76.4 69.9
23 124 123 21 78 70 14 75 2 22 27 579
16 147 237 25 72 31 8 31 5 9 12 593
42 268 542 43 313 75 47 528 34 54 68 2014
16 187 1089 85 272 86 34 273 31 42 46 2161
41 110 652 33 401 55 89 1181 141 60 78 2841
6 76 1150 86 311 68 75 770 66 96 90 2794
0.58 0.23 0.37 0.58 0.28 0.67 0.14 0.26 0.59 0.14 0.09
HWE, HardyeWeinberg equilibrium. The p value of HWE determined by c2 test or Fisher’s exact test among controls.
omitted one at a time and differential estimates were calculated for the remaining research (24). Publication bias was determined by visual inspection of the funnel plot accompanied with the Egger’s linear regression and Begg-Mazemdar tests. An asymmetric plot was verified by determining whether the intercept deviated remarkably from zero in a regression of the standardized effect estimates against their precision (16); p !0.05 was considered to be significant. A cumulative meta-analysis was done to investigate whether the first published research influenced the subsequent publications as well as the evolution of the combined estimates over time in the light of the ascending date of publications. HWE was tested with the chi-square test or Fisher’s exact test based on a Web program (http://ihg2. helmholtz-muenchen.de/cgi-bin/hw/hwa1.pl). Data processing and statistical analyses were performed using Review Manager software 5.0 (Oxford, England) and Stata 11.0 (Stata Corporation, College Station, TX); all p values are two-tailed.
Results Flow of Included Studies A flowchart detailing the selection process is shown in Figure 1. The primary search generated 131 potentially relevant articles, of which 15 examined the association between the C1019T polymorphism and CAD risk and were included in the provisional data set. Among these, four studies were excluded for the following reasons: two Japanese studies (25,26) and one Chinese study overlapped with other studies containing larger cohorts (14,27) and the genotyping frequency of the control population in the Wong et al. (28) study departed from HWE (PHWE 5 0.03). Ultimately, a total of 11 studies had adequate information and
satisfied the inclusion criteria. The 11 included studies were published between 2001 and 2013 with seven studies having been conducted in an Asian population (14,27,29) and four studies in a Caucasian population (30e33). Seven of the studies were written in English (14,27,29e33) and four were written in Chinese. A retrospective case-control design had been utilized in all of the included studies. CAD was considered to be the primary endpoint in five studies (27,29), MI was identified as the endpoint in four studies (14,31,33) and ACS was the endpoint in two studies (30,32). Controls were recruited from certain population groups in two studies (30,31) and nine studies used H-B controls (14,27,29,32,33). Furthermore, two studies matched CAD patients and controls for age and gender (29,32), whereas the other nine studies did not mention matching for age and gender (14,27,30,31,33). Quantitative Data Synthesis Selected clinical characteristics from the eligible studies are shown in Table 1. A total of 5535 CAD patients and 5626 controls were analyzed. The pooled overall frequency of the 1019T allele was 29.2% in CAD cases and 30.1% in controls. The 1019T allele frequency among Asians (26.8 in cases vs. 26.6% in controls) was slightly lower than that among Caucasians (32.4 in cases vs. 32.2% in controls). The allele and genotype frequencies of the C1019T polymorphism for each study that corresponded to HWE (PHWEO0.05) are also listed in Table 2. The correlation of the C1019T polymorphism with CAD risk was calculated across the four genetic models. Comparison of the 1019T allele with the 1019C allele yielded a moderate increased risk for CAD, although this was not significant (allele comparison: p 5 0.34, OR 5 1.11, 95% CI 0.90e1.36; dominant model: p 5 0.44, OR 5 1.09, 95% CI 0.87e1.37; recessive model: p 5 0.37, OR 5 1.18, 95% CI 0.83e1.68; homozygote comparison:
26
Wu et al./ Archives of Medical Research 45 (2014) 21e30
Table 3. Summary estimates for ORs and 95% CI in different subgroups under various genetic contrasts Genotype contrasts Total studies Allele comparison (T vs. C) Dominant model (CTþTT vs. CC) Recessive model (TT vs. CTþCC) Homozygote comparison (TT vs. CC) Ethnicity Allele comparison Dominant model Recessive model Homozygote comparison Endpoint Allele comparison
Dominant model
Recessive model
Homozygote comparison
Study design Allele comparison Dominant model Recessive model Homozygote comparison Population source Allele comparison Dominant model Recessive model Homozygote comparison
Study population
Study number, (case/control), n (n/n)
Pheterogeneity
I2 , %
pa
OR
95% CI
11 11 11 11
(5535/5626) (5535/5626) (5535/5626) (5535/5626)
0.00 0.00 0.00 0.00
88.2 82.1 82.5 83.4
0.34 0.44 0.37 0.36
1.11 1.09 1.18 1.21
0.90e1.36 0.87e1.37 0.83e1.68 0.81e1.80
Caucasian Asian Caucasian Asian Caucasian Asian Caucasian Asian
4 7 4 7 4 7 4 7
(2507/3590) (3028/2036) (2507/3590) (3028/2036) (2507/3590) (3028/2036) (2507/3590) (3028/2036)
0.00 0.00 0.02 0.00 0.00 0.00 0.00 0.00
87.0 89.9 68.5 84.1 83.7 83.1 84.2 85.5
0.26 0.81 0.67 0.79 0.19 0.93 0.23 0.93
1.17 1.04 1.05 1.06 1.38 1.03 1.38 1.03
0.89e1.53 0.74e1.48 0.83e1.34 0.72e1.55 0.85e2.22 0.57e1.84 0.82e2.32 0.51e2.07
CAD MI ACS CAD MI ACS CAD MI ACS CAD MI ACS
5 4 2 5 4 2 5 4 2 5 4 2
(1108/815) (2217/1602) (2210/3209) (1108/815) (2217/1602) (2210/3209) (1108/815) (2217/1602) (2210/3209) (1108/815) (2217/1602) (2210/3209)
0.00 0.02 0.81 0.00 0.12 0.98 0.00 0.23 0.65 0.00 0.19 0.69
86.8 68.5 0.0 82.7 49.0 0.0 76.1 29.7 0.0 82.3 37.8 0.0
0.49 0.00 0.06 0.48 0.00 0.03 0.41 0.00 0.56 0.41 0.00 0.25
0.86 1.59 0.92 0.82 1.55 0.88 0.76 2.05 0.94 0.69 2.28 0.89
0.55e1.33 1.24e2.03 0.84e1.00 0.48e1.41 1.19e2.00 0.79e0.99 0.39e1.48 1.47e2.86 0.78e1.14 0.29e1.67 1.56e3.34 0.73e1.09
Matched Not mentioned matched Not mentioned Matched Not mentioned Matched Not mentioned
2 9 2 9 2 9 2 9
(988/752) (4547/4874) (988/752) (4547/4874) (988/752) (4547/4874) (988/752) (4547/4874)
0.01 0.00 0.03 0.00 0.11 0.00 0.08 0.00
83.6 88.7 77.8 81.8 61.1 84.9 66.5 85.1
0.23 0.10 0.21 0.13 0.38 0.20 0.33 0.17
0.69 1.22 0.68 1.22 0.57 1.31 0.50 1.38
0.37e1.27 0.97e1.54 0.37e1.25 0.95e1.58 0.16e2.01 0.87e1.98 0.12e2.03 0.87e2.20
P-B H-B P-B H-B P-B H-B P-B H-B
3 8 3 8 3 8 3 8
(1696/2940) (3839/2686) (1696/2940) (3839/2686) (1696/2940) (3839/2686) (1696/2940) (3839/2686)
0.00 0.00 0.01 0.00 0.00 0.00 0.02 0.00
90.6 88.9 78.0 84.0 87.3 80.3 88.0 83.4
0.24 0.85 0.38 0.82 0.16 0.97 0.19 0.94
1.32 1.03 1.21 1.04 1.65 1.01 1.71 1.02
0.84e2.07 0.78e1.36 0.79e1.87 0.76e1.42 0.82e3.34 0.64e1.59 0.76e3.83 0.59e1.76
P-B, population-based; H-B, hospital-based; CAD, coronary artery disease; MI, myocardial infarction; ACS, acute coronary syndrome. a Test for overall effect.
p 5 0.36, OR 5 1.21, 95% CI 0.81e1.80). However, there was evidence of substantial between-study heterogeneity across all of the genotypic models (Pheterogeneity !0.001). The summary results of the meta-analysis are listed in Table 3. Sensitivity Analysis A sensitivity analysis was done to determine which study, if any, was the source of the greatest between-study heterogeneity. No individual study was identified as the
cause for the heterogeneity and the ORs in total analysis (Figure 2). Cumulative Analysis and Publication Bias There was no dramatic evidence indicating that the result of the first published study triggered ensuing replication found in the cumulative meta-analysis (data not shown). In addition, Figure 3 revealed that the funnel plot in which the log of the OR of CAD under the allele comparison was plotted against the standard error of the log of the OR in each
Connexin37 Gene C1019T Polymorphism and Risk of Coronary Artery Disease
Figure 2. Sensitivity analysis of the overall odds ratio on the association between the Cx37 C1019T polymorphism and CAD. Results were computed by omitting each individual study in turn. The two ends of the horizontal lines represent the 95% confidence interval.
individual study. The funnel plot of the total result was symmetric and the results of the Egger’s and Begg-Mazemdar tests suggested an absence of publication bias ( p 5 0.47 for Egger’s test and p 5 0.88 for Begg-Mazemdar test). Subgroup Analysis Subgroup analyses were done in which groups with homogeneous characteristics such as ethnicity, endpoint type, population source and study design were analyzed in order to identify the source of the heterogeneity that was present in the overall estimate. Initially, data were stratified into two groups according to race: Caucasian (four studies that recruited 2507 cases and 3509 controls) and Asian (seven studies that recruited 3028 cases and 2036 controls). The ORs of the C1019T polymorphism were slightly higher in the Caucasian group (allele comparison: p 5 0.26, OR 5 1.17, 95% CI 0.89e1.53) compared with the Asian group (allele comparison: p 5 0.81, OR 5 1.04, 95% CI 0.74e1.48) although this was not significant and the ORs
Figure 3. Funnel plot analysis to detect publication bias for allele comparison (T vs. C) of the C1019T polymorphism.
27
were similar between these two subgroups under the dominant model. The probability of different disease endpoints affecting the variability of risk estimates was evaluated by separating the data by CAD, MI and ACS. We observed that the magnitude of the relationship in the MI studies was remarkably reinforced and that the 1019T allele conferred a significantly increased risk of MI (allele comparison: p !0.001, OR 5 1.59, 95% CI 1.24e2.03; dominant model: p !0.001, OR 5 1.55, 95% CI 1.19e2.00) (Figure 4). Interestingly, the effect size was reversed in the CAD (allele comparison: p 5 0.49, OR 5 0.86, 95% CI 0.55e1.33) and in the ACS groups (allele comparison: p 5 0.06, OR 5 0.92, 95% CI 0.84e1.00) and there was no attainable significance except a marginal decreased risk of ACS was noted under the dominant model ( p 5 0.03, OR 5 0.88, 95% CI 0.79e0.99). Further subgroup analyses by population source and study design indicated that there was no striking correlation between the C1019T polymorphism and CAD risk in any genetic model (Table 3). Meta-regression Analysis A meta-regression analysis was performed to further investigate the pre-defined underlying sources of heterogeneity. Multiple study-level covariates such as ethnicity, population source, case definition, study-design, and clinical characteristics of the total population (age, gender, BMI, the average levels of lipid profiles, as well as the proportion of HTN, DM and smoking status) were taken into account. The result indicated that DM contributed largely to the heterogeneity ( p 5 0.014). Namely, the CAD risk of the 1019T allele increased in populations with a high DM rate.
Discussion The relationship between Cx genes and cardiovascular diseases has been unclear in previous studies. A genome-wide association study of 13,372 African Americans documented that an intergenic region downstream of the Cx43 gene was associated with variation in resting heart rate (34). Another genome-wide linkage analysis identified that the diseasesusceptibility locus for MI on chromosome 1p34-36 contained the Cx37 gene (13). However, two large-scale association studies yielded inconsistent results (14,32). We conducted a meta-analysis of publicly available data to comprehensively investigate the relationship between Cx genes and CAD. To the best of our knowledge, this study including 5535 cases and 5626 controls is the first meta-analysis that sheds light on this issue. Although some statistical biases cannot be completely excluded, our study suggests that 1019T carriers have a moderately increased risk of developing CAD. Considering the fact that distinguishable heterogeneity is present and inevitable in any disease identification strategy (35), we performed a panel of subgroup analyses to
28
Wu et al./ Archives of Medical Research 45 (2014) 21e30
Figure 4. Meta-analysis for the association between the C1019T polymorphism and myocardial infarction risk under the allele comparison (T vs. C, Figure 4A), the dominant model (CTþTT vs. CC, Figure 4B), the recessive model (TT vs. CTþCC, Figure 4C) and the homozygote comparison (TT vs. CC, Figure 4D). The 1019T carriers had a significantly increased risk of myocardial infarction. (A color figure can be found in the online version of this article.)
investigate the potential origins of between-study heterogeneity. We hypothesized that the C1019T polymorphism probably had different roles in different racial populations. The risk estimate of the C1019T polymorphism in Caucasian group differed from that in Asian group. Although the 1019T allele frequency between the two ethnic groups was similar, we speculated that the pleiotropic effect of the C1019T polymorphism, due to various genetic ancestral backgrounds, could be an explanation. However, our metaanalysis was restricted to a relatively small sample size in each group. Thus, it is vital that our results be confirmed in larger well-designed studies. Additionally, the inconsistent results from published association studies might be partly due to comparing different study endpoints (e.g., CAD vs. MI). Although the pathological processes of CAD and MI involve many of the same risk factors, these two phenomena are quite different when taking into account features of the atherosclerotic plaque. Acute coronary thrombosis, due to a vulnerable or unstable atherosclerotic plaque, is the key process in MI (33,34). Plaque stability is partially dependent on the activities of
macrophages that secrete pro-inflammatory cytokines and matrix metalloproteinases that make the plaque unstable. The Cx37 C1019T polymorphism encodes hemichannels that distinctively regulate cell adhesiveness. The amino acid alteration at codon 319 probably alters the biophysical properties of the Cx37 hemichannels, thereby affecting ATPdependent cell adhesion. It has been documented that the Cx 37-319S protein, which is coded by the 1019T polymorphism, may enhance the adhesion of macrophages compared with the Cx37-319P protein (12). In contrast, the Cx37-319P allows macrophages to exit the affected area and prevents excessive monocyte recruitment, exhibiting a ‘‘protector’’ role in atherosclerosis (11). Although it is still unclear whether the Cx37 C1019T polymorphism has any functional significance, our results support previous evidence that the Cx37 C1019T polymorphism is associated with the occurrence of MI. In addition to genetic risk factors, environmental influences are another likely explanation for between-study heterogeneity. The significant heterogeneity in the overall and subgroup analyses implies that there may be some clinical
Connexin37 Gene C1019T Polymorphism and Risk of Coronary Artery Disease
characteristics such as age, gender, lifestyle or other related diseases that need to occur before there is any effect or a substantial effect on effect size. We did identify that the proportion of DM had a significant effect on the effect size and the meta regression analysis found a tendency for increased CAD risk in populations with a higher incidence of DM. Increased glucose levels, as happen with diabetes, strengthen monocyte adhesion to endothelial cells via increased adhesion molecules and accelerate atherosclerosis (36). Although meta-regression suggests an ecological correlation rather than a causal inference, our results suggest that the Cx37 C1019T polymorphism and CAD risk are potentially correlated with DM. Further genetic association studies examining the influence of DM are required to accurately quantify the effect size. Despite the fact that our study was partially based on a relatively large sample size and conformed to HWE (37), there were some technical limitations (38). Possible publication bias could not be completely ruled out although the Egger’s test and the funnel plot did not indicate bias. Small negative results tend to be rejected for publication and the ‘‘grey’’ literatures (articles in languages other than English and Chinese) are likely not included. Funnel plot probes whether small studies with little precision give different results than larger studies with greater precision. Asymmetry in the funnel plot may occur as a result of an essential difference between the small and larger studies that arises from inherent between-study heterogeneity (39). We could not completely exclude the probability that small negative or unpublished studies were missing from the plot. In addition, we focused on a single nucleotide polymorphism and did not consider the possibility of gene-gene or haplotype-based effects (40,41). It is likely that the C1019T polymorphism moderately contributes to the risk of CAD patients and additional research is required to investigate whether this polymorphism, combined with other risk factors, will enhance the risk. In conclusion, the combined results of independent association studies suggest that the Cx37 C1019T polymorphism may be a modest risk factor for CAD, especially in MI patients. Moderate to large heterogeneity was identified between the studies and the proportion of DM was a statistically significant source for the observed variation. Our meta-analysis emphasized the necessity of taking genegene and gene-environment interactions into account when trying to interpret and combine data. More population surveys and function researches may facilitate understanding of the true relationship. Acknowledgments This work was supported by the Science and Technology Fund of Shanghai Jiao Tong University School of Medicine (11XJ21001) and the National Natural Science Foundation of China (81201839 and 81370401).
29
References 1. Gensini GF, Comeglio M, Colella A. Classical risk factors and emerging elements in the risk profile for coronary artery disease. Eur Heart J 1998;19(suppl A):A53eA61. 2. Topol EJ, Smith J, Plow EF, et al. Genetic susceptibility to myocardial infarction and coronary artery disease. Hum Mol Genet 2006;15(Spec No 2):R117eR123. 3. Crea F, Liuzzo G. Pathogenesis of acute coronary syndromes. J Am Coll Cardiol 2013;61:1e11. 4. Boerma M, Forsberg L, Van Zeijl L, et al. A genetic polymorphism in connexin 37 as a prognostic marker for atherosclerotic plaque development. J Intern Med 1999;246:211e218. 5. Koval M. Pathways and control of connexin oligomerization. Trends Cell Biol 2006;16:159e166. 6. Laird DW. Life cycle of connexins in health and disease. Biochem J 2006;394:527e543. 7. Saez JC, Berthoud VM, Branes MC, et al. Plasma membrane channels formed by connexins: their regulation and functions. Physiol Rev 2003;83:1359e1400. 8. Saez JC, Berthoud VM, Moreno AP, et al. Gap junctions. Multiplicity of controls in differentiated and undifferentiated cells and possible functional implications. Adv Second Messenger Phosphoprotein Res 1993;27:163e198. 9. Kwak BR, Mulhaupt F, Veillard N, et al. Altered pattern of vascular connexin expression in atherosclerotic plaques. Arterioscler Thromb Vasc Biol 2002;22:225e230. 10. Wong CW, Christen T, Kwak BR. Connexins in leukocytes: shuttling messages? Cardiovasc Res 2004;62:357e367. 11. Wong CW, Christen T, Roth I, et al. Connexin37 protects against atherosclerosis by regulating monocyte adhesion. Nat Med 2006;12:950e954. 12. Chanson M, Kwak BR. Connexin37: a potential modifier gene of inflammatory disease. J Mol Med (Berl) 2007;85:787e795. 13. Wang Q, Rao S, Shen GQ, et al. Premature myocardial infarction novel susceptibility locus on chromosome 1P34-36 identified by genomewide linkage analysis. Am J Hum Genet 2004;74:262e271. 14. Yamada Y, Izawa H, Ichihara S, et al. Prediction of the risk of myocardial infarction from polymorphisms in candidate genes. N Engl J Med 2002;347:1916e1923. 15. Lanfear DE, Jones PG, Marsh S,et al. Connexin37 (GJA4) genotype predicts survival after an acute coronary syndrome. Am Heart J 2007;154:561e566. 16. Egger M, Davey Smith G, Schneider M, et al. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997;315:629e634. 17. Anderson JL, Adams CD, Antman EM, et al. 2011 ACCF/AHA Focused Update Incorporated Into the ACC/AHA 2007 Guidelines for the Management of Patients With Unstable Angina/Non-STElevation Myocardial Infarction: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. Circulation 2011;123:e426ee579. 18. Joensen AM, Jensen MK, Overvad K, et al. Predictive values of acute coronary syndrome discharge diagnoses differed in the Danish National Patient Registry. J Clin Epidemiol 2009;62:188e194. 19. Wright RS, Anderson JL, Adams CD, et al. 2011 ACCF/AHA focused update incorporated into the ACC/AHA 2007 Guidelines for the Management of Patients with Unstable Angina/Non-ST-Elevation Myocardial Infarction: a report of the American College of Cardiology Foundation/ American Heart Association Task Force on Practice Guidelines developed in collaboration with the American Academy of Family Physicians, Society for Cardiovascular Angiography and Interventions, and the Society of Thoracic Surgeons. J Am Coll Cardiol 2011;57:e215ee367. 20. Cohn LD, Becker BJ. How meta-analysis increases statistical power. Psychol Methods 2003;8:243e253. 21. Lau J, Ioannidis JP, Schmid CH. Quantitative synthesis in systematic reviews. Ann Intern Med 1997;127:820e826. 22. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21:1539e1558.
30
Wu et al./ Archives of Medical Research 45 (2014) 21e30
23. Higgins JP, Thompson SG, Deeks JJ, et al. Measuring inconsistency in meta-analyses. BMJ 2003;327:557e560. 24. Biros E, Norman PE, Jones GT, et al. Meta-analysis of the association between single nucleotide polymorphisms in TGF-beta receptor genes and abdominal aortic aneurysm. Atherosclerosis 2011;219:218e223. 25. Yamada Y, Ichihara S, Izawa H, et al. Genetic risk for coronary artery disease in individuals with or without type 2 diabetes. Mol Genet Metab 2004;81:282e290. 26. Hirashiki A, Yamada Y, Murase Y, et al. Association of gene polymorphisms with coronary artery disease in low- or high-risk subjects defined by conventional risk factors. J Am Coll Cardiol 2003;42:1429e1437. 27. Han Y, Xi S, Zhang X, et al. Association of connexin 37 gene polymorphisms with risk of coronary artery disease in northern Han Chinese. Cardiology 2008;110:260e265. 28. Wong CW, Christen T, Pfenniger A, et al. Do allelic variants of the connexin37 1019 gene polymorphism differentially predict for coronary artery disease and myocardial infarction? Atherosclerosis 2007; 191:355e361. 29. Yeh HI, Chou Y, Liu HF, et al. Connexin37 gene polymorphism and coronary artery disease in Taiwan. Int J Cardiol 2001;81:251e255. 30. Hubacek JA, Stanek V, Gebauerova M, et al. Lack of an association between connexin-37, stromelysin-1, plasminogen activator-inhibitor type 1 and lymphotoxin-alpha genes and acute coronary syndrome in Czech Caucasians. Exp Clin Cardiol 2010;15:e52ee56. 31. Listi F, Candore G, Lio D, et al. Association between C1019T polymorphism of connexin37 and acute myocardial infarction: a study in patients from Sicily. Int J Cardiol 2005;102:269e271. 32. Morgan TM, Krumholz HM, Lifton RP, et al. Nonvalidation of reported genetic risk factors for acute coronary syndrome in a largescale replication study. JAMA 2007;297:1551e1561.
33. Seifi M, Fallah S, Ghasemi A, et al. Mutations of the connexin 37 and 40 gap-junction genes in patients with acute myocardial infarction. Clin Lab 2013;59:343e348. 34. Deo R, Nalls MA, Avery CL, et al. Common genetic variation near the connexin-43 gene is associated with resting heart rate in African Americans: a genome-wide association study of 13,372 participants. Heart Rhythm 2013;10:401e408. 35. Hemminki K, Lorenzo Bermejo J, Forsti A. The balance between heritable and environmental aetiology of human disease. Nat Rev Genet 2006;7:958e965. 36. Piga R, Naito Y, Kokura S, et al. Short-term high glucose exposure induces monocyte-endothelial cells adhesion and transmigration by increasing VCAM-1 and MCP-1 expression in human aortic endothelial cells. Atherosclerosis 2007;193:328e334. 37. Munafo MR, Flint J. Meta-analysis of genetic association studies. Trends Genet 2004;20:439e444. 38. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000;283:2008e2012. 39. Lau J, Ioannidis JP, Terrin N, et al. The case of the misleading funnel plot. BMJ 2006;333:597e600. 40. Wu Z, Lou Y, Jin W, et al. The C161T polymorphism in the peroxisome proliferator-activated receptor gamma gene (PPARgamma) is associated with risk of coronary artery disease: a meta-analysis. Mol Biol Rep 2013;40:3101e3112. 41. Wu Z, Lou Y, Jin W, et al. The Pro12Ala polymorphism in the peroxisome proliferator-activated receptor gamma-2 gene (PPARgamma2) is associated with increased risk of coronary artery disease: a metaanalysis. PLoS One 2012;7:e53105.
ORIGINAL ARTICLE
The Connexin37 Gene C1019T Polymorphism and Risk of Coronary Artery Disease: A Meta-analysis Zhijun Wu,a Yuqing Lou,b Wei Jin,a Yan Liu,a Lin Lu,a Qiujing Chen,a and Ruiyan Zhanga a
Department of Cardiology, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, PR China Department of Pulmonary Diseases, Shanghai Chest Hospital, Shanghai Jiao Tong University, Shanghai, PR China
b
Received for publication July 3, 2013; accepted October 18, 2013 (ARCMED-D-13-00355).
Background and Aims. Mounting data have emerged suggesting that the Connexin37 C1019T polymorphism increases susceptibility to coronary artery disease (CAD). However, previous studies yielded conflicting results. In the current study, a comprehensive meta-analysis was performed to investigate whether the C1019T polymorphism is associated with CAD risk. Methods. A total of 11 studies examining the C1019T polymorphism and CAD were identified using MEDLINE, Embase, CNKI, Wanfang and CBM, in which 5535 CAD patients and 5626 controls were analyzed. A random-effects model was used to calculate odd ratios and confidence intervals, while addressing between-study heterogeneity. Publication bias was weighed using the Egger’s test, Begg-Mazemdar test and funnel plot. Results. In genetic models with striking heterogeneity, the risk of CAD was not associated with the C1019T polymorphism (allele comparison: p 5 0.34, OR 5 1.11, 95% CI 0.90e1.36). Stratification by disease endpoints indicated that the 1019T allele was significantly associated with myocardial infarction (MI) (allele comparison: p !0.001, OR 5 1.59, 95% CI 1.24e2.03). Further meta-regression analysis indicated that a large proportion of heterogeneity was probably due to the varying proportions of diabetes mellitus (DM) across studies ( p 5 0.014). Conclusions. Our results indicated that the C1019T polymorphism may be a moderate risk factor for MI and that DM was likely a potential source of between-study heterogeneity. Ó 2014 IMSS. Published by Elsevier Inc. Key Words: Connexin, Coronary artery disease, Myocardial infarction, Polymorphism Metaanalysis.
Introduction A considerable portion of coronary artery disease (CAD) and myocardial infarction (MI) events cannot be characterized by the traditional risk factors. Diseases and conditions such as hypertension (HTN), diabetes (DM) and hyperlipidemia, or other characteristics and habits such as being overweight and smoking are generally regarded as contributors to cardiovascular diseases (1). In addition, a broad-spectrum of Address reprint requests to: Ruiyan Zhang, MD, PhD, Department of Cardiology, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, No 197, Ruijin Er Road, Shanghai, 200025, PR China; Phone: (þ86) 21-64370045; FAX: (þ86) 21-64457177; E-mail: zhangruiyan@ 263.net
underlying pathological processes, ranging from lipoprotein deposits in the arterial intima to blood thrombosis formation due to plaque rupture, have been associated with cardiovascular disease (2). Among the major pathways, endothelial integrity and persistent inflammation appear to be key factors that play roles in the biological etiology of CAD and MI (3). The endothelial barrier is supported by gap- junctional proteins and prevents untoward diapedesis and penetration of low-density lipoprotein cholesterol (4). Connexins are members of a family of endothelial gapjunctional proteins and are expressed in the vascular wall. Six connexin subunits form a hemi-channel called a connexon in the plasma membrane. Connexons dock with their counterparts in the plasma membrane of an adjacent cell
0188-4409/$ - see front matter. Copyright Ó 2014 IMSS. Published by Elsevier Inc. http://dx.doi.org/10.1016/j.arcmed.2013.12.001
22
Wu et al./ Archives of Medical Research 45 (2014) 21e30
and assemble an intercellular channel called a gap junction (5,6), Gap junctions contribute to the maintenance of cell growth and differentiation (7). In the human genome, the connexin gene family is comprised of 21 genes. Different connexins are named according to the molecular weight that is predicted by their respective cDNAs (8). Connexin 37 (Cx 37) has a unique expression in endothelial and macrophage cells and plays a key role in atherogenesis (9,10). Deletion of Cx37 accelerates atherosclerosis in Cx37/ApoE/mice compared with the Cx37þ/þApoE/ controls, suggesting that Cx37 may exert an anti-atherosclerosis effect on ApoE/ mice (11). Additionally, there is a higher proportion of Cx37deficient monocytes and macrophages among leukocytes that are related to atherosclerotic plaques (12). The human Cx37 gene is located at chromosome 1p35.1 (13). One of the most common polymorphisms of the Cx37 gene is a cytosine-to-thymine replacement at position 1019 resulting in a substitution of serine for proline at codon 319 (P319S) in the regulatory C-terminus of Cx37 protein. The Cx37 polymorphism C1019T has been identified as a prognostic marker for atherothrombosis (14,15). The C1019T polymorphism has been significantly associated with a 3-year survival after acute coronary syndrome (ACS) (15). Therefore, it is of added interest to identify whether the C1019T polymorphism is related to the risk of CAD or MI. However, published association studies have reported either inconclusive or conflicting results. Generally speaking, insufficient sample size, marked racial and ethnic differences, unadjusted environmental confounders and genetic background noise are likely at least partially responsible for the current discordance among results. Meta analysis is an ideal tool to combine paradoxical results from different studies (16) in order to estimate the major effect with enhanced precision. In the current study, a meta-analysis is done incorporating the available data to date, to investigate whether the Cx37 C1019T polymorphism is a risk factor for the CAD. Between-study heterogeneity and publication bias are also addressed.
Materials and Methods Search Strategy A literature search strategy was undertaken using PubMed/ MEDLINE, Embase, the China National Knowledge Infrastructure (CNKI), Wanfang and the Chinese Biological Medicine (CBM) electronic databases up to June 2013. Combinations of the following key words were used as medical subject heading (MeSH) terms and text words: ‘C1019T’ or ‘Pro319Ser’, ‘connexin 37’, ‘gap junction protein alpha-4’ or ‘GJA4’ and ‘coronary’ or ‘ischemic heart disease’ or ‘myocardial infarction’ or ‘atherosclerosis’. The full texts of potentially appropriate articles were scrutinized to decide whether the data would be examined further.
The bibliographies of relevant studies, reviews and previous meta-analyses were reviewed to identify additional studies that were not indexed by the initial searches. Hand searches and the related search functions in PubMed were also used. If the allele or genotype frequencies were incompletely reported, the corresponding authors we were contacted to obtain the raw data. Selection Criteria Studies related to the association of the Cx37 C1019T polymorphism with CAD risk were provisionally included if (1) they were published in English or Chinese journals or their supplements; (2) they provided sufficient information on the C1019T polymorphism and studies without controls were excluded; (3) the genotype frequency of the control population conformed to Hardy-Weinberg Equilibrium (HWE). There were no restrictions on ethnicity or country in which the studies were conducted. When an article involved more than one subpopulation, each population was considered as a separate study. When multiple studies were derived from the same population, only the largest scale study was included to avoid double calculation of the odd ratios (ORs). The diagnosis of MI was done on the basis of typical electrocardiographic changes (both ST-segment elevation and non-ST-segment elevation changes) and a positive troponin blood test (17). A diagnosis of acute coronary syndrome (ACS) was based on unstable angina pectoris, as well as acute and subacute MI (18). CAD was defined as a stenosis diameter of $50% in any of the major coronary arteries or in the left main trunk as documented by coronary angiography (19). Data Extraction Two authors (ZW and YL) independently obtained the full texts of the eligible articles based on the titles and abstracts. The necessary data was extracted from the selected studies and entered into separate databases in duplicate. Any disagreements were resolved at a consensus meeting. Data were collated according to different cohorts: first author’s name, publication year, ethnicity, geographical location, study design, population source, disease outcome, sample size and clinical characteristics (e.g., age, body mass index [BMI], circulating lipid profiles, the proportion of males, hypertension [HTN], diabetes mellitus [DM] and smoking status). The C1019T genotype frequency and its consistency with HWE were also collected for both patients and controls. Continuous variables are manifested as mean standard deviation (SD) or median (5th and 95th percentiles). Statistical Analysis Allele frequencies of the C1019T polymorphism were determined by the allele counting method. Data were combined
Connexin37 Gene C1019T Polymorphism and Risk of Coronary Artery Disease
23
Figure 1. Flow diagram of the search strategy and study selection for the meta-analysis.
and reported as weighted ORs corresponding to the 95% confidence intervals (CIs) of the association between the C1019T polymorphism and CAD risk. Four genetic models were used: allele comparison (T vs. C), the dominant genetic model (CTþTT vs. CC), the recessive genetic model (TT vs. CTþCC) and homozygote comparison (TT vs. CC). A random-effects model using DerSimonian & Laird method was used to estimate the individual effect size and to modulate study weights based on in-study variance. The method does not account for heterogeneity, thus the Mantel-Haenszel model was used to assess heterogeneity (20). In addition, Cochran’s chi-square based Q statistical test was used to evaluate within- and between-study heterogeneity in which p !0.10 was regarded as significant (21). The I2 statistic was used to evaluate the degree of inconsistency in the studies or, in other words, to estimate the percentage of between-study variability due to heterogeneity
rather than due to chance (22,23). The I2 ranged from 0e100% where a value of 0% indicated no observed heterogeneity and increasing values indicated increasing degrees of heterogeneity. Pooled OR significance was evaluated by the Z test and p !0.05 was considered to be significant. The pre-specified characteristics for evaluation of interstudy heterogeneity sources were as follows: ethnicity (Asian and Caucasian), endpoints (CAD, ACS and MI), study design (matched and not reported), and population source (hospital-based [H-B] and population-based [P-B]). Additionally, a meta-regression was done to evaluate the extent to which different covariates, including the characteristics mentioned above, explained the genetic heterogeneity among the individual ORs. Sensitivity analyses were performed to identify the most influential studies that could contribute to the possibility of biased effects estimation. Each study was successive
24
Table 1. Baseline characteristics of all qualified studies included in the meta-analysis
Ethnicity
Geographic location Source Endpoint
Year
Feng W
2010 Asian
China
H-B
CAD
Han YL
2007 Asian
China
H-B
CAD
Czekh
P-B
ACS
Hubacek JA 2010 Caucasian
Study design
Status
Not mentioned Cases Controls Not mentioned Cases Controls Not mentioned Cases Controls
Listi F
2005 Caucasian
Morgan TM 2007 Caucasian
Italy
P-B
MI
US
H-B
ACS
Not mentioned Cases Controls Matched Cases Controls
Seifi M
2013 Caucasian
Iran
P-B
MI
Xie XJ
2006 Asian
China
H-B
CAD
Yamada Y
2002 Asian
Japan
H-B
MI
Yeh HI
2001 Asian
China
H-B
CAD
Zhang CL
2009 Asian
China
H-B
MI
Zhang SR
2007 Asian
China
H-B
CAD
Not mentioned Cases Controls Not mentioned Cases Controls Not mentioned Cases Controls matched Cases Controls Not mentioned Cases Controls Not mentioned Cases Controls
Age, year 64.4 11.4 57.0 10.4 59.3 10.6 52.9 10.4 55.2 7.5 (M) 62.6 8.5 (F) 49.0 10.8 (M) 48.6 10.6 (F) 40.0 (20.0e46.0) 39.0 (20.0e55.0) 60.7 12.5 (M) 63.1 13.2 (F) 60.0 12.1 (M) 61.8 12.8 (F) 58.0 13.0 57.0 13.0 61.0 10.0 60.0 10.0 62.1 10.1 62.0 10.4 61.0 11.0 60.0 11.0 57.2 12.6 50.6 11.1 68.0 10.9 66.7 8.6
Gender, M (%) HTN, % DM, % Smoke, % 50.0 47.4 77.3 59.0 72.6
20.8 23.7 59.4 46.6 55.5
15.1 15.8 17.7 12.9 39.0
37.7 26.3 40.6 30.2 73.0
46.5
36.7
7.8
26.7
100.0 100.0 67.8
31.0 60.1
17.0 23.8
-a 33.1
60.6
51.2
11.9
13.2
64.0 64.3 81.3 61.5 100.0 100.0 67.8 68.6 72.8 60.5 78.6 58.1
41.3 37.6 47.0 57.3 80.5 55.7
34.7 16.2 23.0 21.0 46.1 17.5
42.7 21.4 57.8 55.2 49.0 47.0 49.7 20.6
BMI, kg/m2 24.2 24.2 25.4 25.0 28.5 28.8 28.2 27.6
29.1 29.9 27.9 27.7 26.3 26.7 25.0 25.0 23.6 23.6 25.2 25.5
24.5 23.5
1.7 1.5 3.0 3.2 4.3 5.6 4.0 5.5 5.5 6.9 5.0 6.9 4.7 4.9 3.0 4.0 2.9 2.7 3.2 3.7 3.7 5.1
TC, mmol/L
4.0 3.9 4.6 4.5 (M) 5.2 (F) 5.4 (M) 5.8 (F) 5.8
(M) (F) (M) (F) 5.1 4.9 5.0 4.6
5.2 5.0 4.8 4.4
0.9 0.8 1.0 0.8 1.2 1.3 1.1 1.2 1.1 1.0 1.1 0.9 1.1 1.2 1.2 1.4 -
Tg, mmol/L
1.3 1.3 2.0 2.0 (M) 2.1 (F) 1.9 (M) 2.0 (F) 1.5
1.9 1.8 1.6 1.5
2.0 1.9 1.6 1.5
0.4 0.5 1.4 1.6 1.5 1.2 1.3 0.8 1.1 1.1 1.2 1.3 1.3 1.3 0.7 0.9 -
(M) (F) (M) (F)
P-B, population-based study; H-B, hospital-based study; CAD, coronary artery disease; MI, myocardial infarction; ACS, acute coronary syndrome; M(%), male (percent); F, female; HTN, hypertension; DM, diabetes mellitus; BMI, body mass index; TC, total cholesterol; Tg, triglyceride. a Data not available; age and BMI are expressed as mean SD (standard deviation) or median (5th and 95th percentiles).
Wu et al./ Archives of Medical Research 45 (2014) 21e30
First author
Connexin37 Gene C1019T Polymorphism and Risk of Coronary Artery Disease
25
Table 2. Sample size, distribution of the C1019T allele frequencies and genotypes among CAD cases and controls, and p value of HWE in controls Sample size
T allele, %
C allele, %
TT genotype
CT genotype
CC genotype
HWE,
First author
Cases
Controls
Cases
Controls
Cases
Controls
Cases
Controls
Cases
Controls
Cases
Controls
p
Feng W Han YL Hubacek JA Listi F Morgan TM Seifi M Xie XJ Yamada Y Yeh HI Zhang CL Zhang SR Total
106 502 1399 97 811 200 150 1784 177 136 173 5535
38 410 2559 196 650 185 117 1074 102 147 148 5626
43.9 51.4 29.9 43.8 29.6 53.8 25.0 19.0 10.7 36.0 35.3 29.2
59.2 58.7 31.6 34.4 31.8 40.0 21.4 15.6 20.1 20.4 23.7 30.1
56.1 48.6 70.1 56.2 70.4 46.3 75.0 81.0 89.3 64.0 64.7 70.8
40.8 41.3 68.4 65.6 68.2 60.0 78.6 84.4 79.9 79.6 76.4 69.9
23 124 123 21 78 70 14 75 2 22 27 579
16 147 237 25 72 31 8 31 5 9 12 593
42 268 542 43 313 75 47 528 34 54 68 2014
16 187 1089 85 272 86 34 273 31 42 46 2161
41 110 652 33 401 55 89 1181 141 60 78 2841
6 76 1150 86 311 68 75 770 66 96 90 2794
0.58 0.23 0.37 0.58 0.28 0.67 0.14 0.26 0.59 0.14 0.09
HWE, HardyeWeinberg equilibrium. The p value of HWE determined by c2 test or Fisher’s exact test among controls.
omitted one at a time and differential estimates were calculated for the remaining research (24). Publication bias was determined by visual inspection of the funnel plot accompanied with the Egger’s linear regression and Begg-Mazemdar tests. An asymmetric plot was verified by determining whether the intercept deviated remarkably from zero in a regression of the standardized effect estimates against their precision (16); p !0.05 was considered to be significant. A cumulative meta-analysis was done to investigate whether the first published research influenced the subsequent publications as well as the evolution of the combined estimates over time in the light of the ascending date of publications. HWE was tested with the chi-square test or Fisher’s exact test based on a Web program (http://ihg2. helmholtz-muenchen.de/cgi-bin/hw/hwa1.pl). Data processing and statistical analyses were performed using Review Manager software 5.0 (Oxford, England) and Stata 11.0 (Stata Corporation, College Station, TX); all p values are two-tailed.
Results Flow of Included Studies A flowchart detailing the selection process is shown in Figure 1. The primary search generated 131 potentially relevant articles, of which 15 examined the association between the C1019T polymorphism and CAD risk and were included in the provisional data set. Among these, four studies were excluded for the following reasons: two Japanese studies (25,26) and one Chinese study overlapped with other studies containing larger cohorts (14,27) and the genotyping frequency of the control population in the Wong et al. (28) study departed from HWE (PHWE 5 0.03). Ultimately, a total of 11 studies had adequate information and
satisfied the inclusion criteria. The 11 included studies were published between 2001 and 2013 with seven studies having been conducted in an Asian population (14,27,29) and four studies in a Caucasian population (30e33). Seven of the studies were written in English (14,27,29e33) and four were written in Chinese. A retrospective case-control design had been utilized in all of the included studies. CAD was considered to be the primary endpoint in five studies (27,29), MI was identified as the endpoint in four studies (14,31,33) and ACS was the endpoint in two studies (30,32). Controls were recruited from certain population groups in two studies (30,31) and nine studies used H-B controls (14,27,29,32,33). Furthermore, two studies matched CAD patients and controls for age and gender (29,32), whereas the other nine studies did not mention matching for age and gender (14,27,30,31,33). Quantitative Data Synthesis Selected clinical characteristics from the eligible studies are shown in Table 1. A total of 5535 CAD patients and 5626 controls were analyzed. The pooled overall frequency of the 1019T allele was 29.2% in CAD cases and 30.1% in controls. The 1019T allele frequency among Asians (26.8 in cases vs. 26.6% in controls) was slightly lower than that among Caucasians (32.4 in cases vs. 32.2% in controls). The allele and genotype frequencies of the C1019T polymorphism for each study that corresponded to HWE (PHWEO0.05) are also listed in Table 2. The correlation of the C1019T polymorphism with CAD risk was calculated across the four genetic models. Comparison of the 1019T allele with the 1019C allele yielded a moderate increased risk for CAD, although this was not significant (allele comparison: p 5 0.34, OR 5 1.11, 95% CI 0.90e1.36; dominant model: p 5 0.44, OR 5 1.09, 95% CI 0.87e1.37; recessive model: p 5 0.37, OR 5 1.18, 95% CI 0.83e1.68; homozygote comparison:
26
Wu et al./ Archives of Medical Research 45 (2014) 21e30
Table 3. Summary estimates for ORs and 95% CI in different subgroups under various genetic contrasts Genotype contrasts Total studies Allele comparison (T vs. C) Dominant model (CTþTT vs. CC) Recessive model (TT vs. CTþCC) Homozygote comparison (TT vs. CC) Ethnicity Allele comparison Dominant model Recessive model Homozygote comparison Endpoint Allele comparison
Dominant model
Recessive model
Homozygote comparison
Study design Allele comparison Dominant model Recessive model Homozygote comparison Population source Allele comparison Dominant model Recessive model Homozygote comparison
Study population
Study number, (case/control), n (n/n)
Pheterogeneity
I2 , %
pa
OR
95% CI
11 11 11 11
(5535/5626) (5535/5626) (5535/5626) (5535/5626)
0.00 0.00 0.00 0.00
88.2 82.1 82.5 83.4
0.34 0.44 0.37 0.36
1.11 1.09 1.18 1.21
0.90e1.36 0.87e1.37 0.83e1.68 0.81e1.80
Caucasian Asian Caucasian Asian Caucasian Asian Caucasian Asian
4 7 4 7 4 7 4 7
(2507/3590) (3028/2036) (2507/3590) (3028/2036) (2507/3590) (3028/2036) (2507/3590) (3028/2036)
0.00 0.00 0.02 0.00 0.00 0.00 0.00 0.00
87.0 89.9 68.5 84.1 83.7 83.1 84.2 85.5
0.26 0.81 0.67 0.79 0.19 0.93 0.23 0.93
1.17 1.04 1.05 1.06 1.38 1.03 1.38 1.03
0.89e1.53 0.74e1.48 0.83e1.34 0.72e1.55 0.85e2.22 0.57e1.84 0.82e2.32 0.51e2.07
CAD MI ACS CAD MI ACS CAD MI ACS CAD MI ACS
5 4 2 5 4 2 5 4 2 5 4 2
(1108/815) (2217/1602) (2210/3209) (1108/815) (2217/1602) (2210/3209) (1108/815) (2217/1602) (2210/3209) (1108/815) (2217/1602) (2210/3209)
0.00 0.02 0.81 0.00 0.12 0.98 0.00 0.23 0.65 0.00 0.19 0.69
86.8 68.5 0.0 82.7 49.0 0.0 76.1 29.7 0.0 82.3 37.8 0.0
0.49 0.00 0.06 0.48 0.00 0.03 0.41 0.00 0.56 0.41 0.00 0.25
0.86 1.59 0.92 0.82 1.55 0.88 0.76 2.05 0.94 0.69 2.28 0.89
0.55e1.33 1.24e2.03 0.84e1.00 0.48e1.41 1.19e2.00 0.79e0.99 0.39e1.48 1.47e2.86 0.78e1.14 0.29e1.67 1.56e3.34 0.73e1.09
Matched Not mentioned matched Not mentioned Matched Not mentioned Matched Not mentioned
2 9 2 9 2 9 2 9
(988/752) (4547/4874) (988/752) (4547/4874) (988/752) (4547/4874) (988/752) (4547/4874)
0.01 0.00 0.03 0.00 0.11 0.00 0.08 0.00
83.6 88.7 77.8 81.8 61.1 84.9 66.5 85.1
0.23 0.10 0.21 0.13 0.38 0.20 0.33 0.17
0.69 1.22 0.68 1.22 0.57 1.31 0.50 1.38
0.37e1.27 0.97e1.54 0.37e1.25 0.95e1.58 0.16e2.01 0.87e1.98 0.12e2.03 0.87e2.20
P-B H-B P-B H-B P-B H-B P-B H-B
3 8 3 8 3 8 3 8
(1696/2940) (3839/2686) (1696/2940) (3839/2686) (1696/2940) (3839/2686) (1696/2940) (3839/2686)
0.00 0.00 0.01 0.00 0.00 0.00 0.02 0.00
90.6 88.9 78.0 84.0 87.3 80.3 88.0 83.4
0.24 0.85 0.38 0.82 0.16 0.97 0.19 0.94
1.32 1.03 1.21 1.04 1.65 1.01 1.71 1.02
0.84e2.07 0.78e1.36 0.79e1.87 0.76e1.42 0.82e3.34 0.64e1.59 0.76e3.83 0.59e1.76
P-B, population-based; H-B, hospital-based; CAD, coronary artery disease; MI, myocardial infarction; ACS, acute coronary syndrome. a Test for overall effect.
p 5 0.36, OR 5 1.21, 95% CI 0.81e1.80). However, there was evidence of substantial between-study heterogeneity across all of the genotypic models (Pheterogeneity !0.001). The summary results of the meta-analysis are listed in Table 3. Sensitivity Analysis A sensitivity analysis was done to determine which study, if any, was the source of the greatest between-study heterogeneity. No individual study was identified as the
cause for the heterogeneity and the ORs in total analysis (Figure 2). Cumulative Analysis and Publication Bias There was no dramatic evidence indicating that the result of the first published study triggered ensuing replication found in the cumulative meta-analysis (data not shown). In addition, Figure 3 revealed that the funnel plot in which the log of the OR of CAD under the allele comparison was plotted against the standard error of the log of the OR in each
Connexin37 Gene C1019T Polymorphism and Risk of Coronary Artery Disease
Figure 2. Sensitivity analysis of the overall odds ratio on the association between the Cx37 C1019T polymorphism and CAD. Results were computed by omitting each individual study in turn. The two ends of the horizontal lines represent the 95% confidence interval.
individual study. The funnel plot of the total result was symmetric and the results of the Egger’s and Begg-Mazemdar tests suggested an absence of publication bias ( p 5 0.47 for Egger’s test and p 5 0.88 for Begg-Mazemdar test). Subgroup Analysis Subgroup analyses were done in which groups with homogeneous characteristics such as ethnicity, endpoint type, population source and study design were analyzed in order to identify the source of the heterogeneity that was present in the overall estimate. Initially, data were stratified into two groups according to race: Caucasian (four studies that recruited 2507 cases and 3509 controls) and Asian (seven studies that recruited 3028 cases and 2036 controls). The ORs of the C1019T polymorphism were slightly higher in the Caucasian group (allele comparison: p 5 0.26, OR 5 1.17, 95% CI 0.89e1.53) compared with the Asian group (allele comparison: p 5 0.81, OR 5 1.04, 95% CI 0.74e1.48) although this was not significant and the ORs
Figure 3. Funnel plot analysis to detect publication bias for allele comparison (T vs. C) of the C1019T polymorphism.
27
were similar between these two subgroups under the dominant model. The probability of different disease endpoints affecting the variability of risk estimates was evaluated by separating the data by CAD, MI and ACS. We observed that the magnitude of the relationship in the MI studies was remarkably reinforced and that the 1019T allele conferred a significantly increased risk of MI (allele comparison: p !0.001, OR 5 1.59, 95% CI 1.24e2.03; dominant model: p !0.001, OR 5 1.55, 95% CI 1.19e2.00) (Figure 4). Interestingly, the effect size was reversed in the CAD (allele comparison: p 5 0.49, OR 5 0.86, 95% CI 0.55e1.33) and in the ACS groups (allele comparison: p 5 0.06, OR 5 0.92, 95% CI 0.84e1.00) and there was no attainable significance except a marginal decreased risk of ACS was noted under the dominant model ( p 5 0.03, OR 5 0.88, 95% CI 0.79e0.99). Further subgroup analyses by population source and study design indicated that there was no striking correlation between the C1019T polymorphism and CAD risk in any genetic model (Table 3). Meta-regression Analysis A meta-regression analysis was performed to further investigate the pre-defined underlying sources of heterogeneity. Multiple study-level covariates such as ethnicity, population source, case definition, study-design, and clinical characteristics of the total population (age, gender, BMI, the average levels of lipid profiles, as well as the proportion of HTN, DM and smoking status) were taken into account. The result indicated that DM contributed largely to the heterogeneity ( p 5 0.014). Namely, the CAD risk of the 1019T allele increased in populations with a high DM rate.
Discussion The relationship between Cx genes and cardiovascular diseases has been unclear in previous studies. A genome-wide association study of 13,372 African Americans documented that an intergenic region downstream of the Cx43 gene was associated with variation in resting heart rate (34). Another genome-wide linkage analysis identified that the diseasesusceptibility locus for MI on chromosome 1p34-36 contained the Cx37 gene (13). However, two large-scale association studies yielded inconsistent results (14,32). We conducted a meta-analysis of publicly available data to comprehensively investigate the relationship between Cx genes and CAD. To the best of our knowledge, this study including 5535 cases and 5626 controls is the first meta-analysis that sheds light on this issue. Although some statistical biases cannot be completely excluded, our study suggests that 1019T carriers have a moderately increased risk of developing CAD. Considering the fact that distinguishable heterogeneity is present and inevitable in any disease identification strategy (35), we performed a panel of subgroup analyses to
28
Wu et al./ Archives of Medical Research 45 (2014) 21e30
Figure 4. Meta-analysis for the association between the C1019T polymorphism and myocardial infarction risk under the allele comparison (T vs. C, Figure 4A), the dominant model (CTþTT vs. CC, Figure 4B), the recessive model (TT vs. CTþCC, Figure 4C) and the homozygote comparison (TT vs. CC, Figure 4D). The 1019T carriers had a significantly increased risk of myocardial infarction. (A color figure can be found in the online version of this article.)
investigate the potential origins of between-study heterogeneity. We hypothesized that the C1019T polymorphism probably had different roles in different racial populations. The risk estimate of the C1019T polymorphism in Caucasian group differed from that in Asian group. Although the 1019T allele frequency between the two ethnic groups was similar, we speculated that the pleiotropic effect of the C1019T polymorphism, due to various genetic ancestral backgrounds, could be an explanation. However, our metaanalysis was restricted to a relatively small sample size in each group. Thus, it is vital that our results be confirmed in larger well-designed studies. Additionally, the inconsistent results from published association studies might be partly due to comparing different study endpoints (e.g., CAD vs. MI). Although the pathological processes of CAD and MI involve many of the same risk factors, these two phenomena are quite different when taking into account features of the atherosclerotic plaque. Acute coronary thrombosis, due to a vulnerable or unstable atherosclerotic plaque, is the key process in MI (33,34). Plaque stability is partially dependent on the activities of
macrophages that secrete pro-inflammatory cytokines and matrix metalloproteinases that make the plaque unstable. The Cx37 C1019T polymorphism encodes hemichannels that distinctively regulate cell adhesiveness. The amino acid alteration at codon 319 probably alters the biophysical properties of the Cx37 hemichannels, thereby affecting ATPdependent cell adhesion. It has been documented that the Cx 37-319S protein, which is coded by the 1019T polymorphism, may enhance the adhesion of macrophages compared with the Cx37-319P protein (12). In contrast, the Cx37-319P allows macrophages to exit the affected area and prevents excessive monocyte recruitment, exhibiting a ‘‘protector’’ role in atherosclerosis (11). Although it is still unclear whether the Cx37 C1019T polymorphism has any functional significance, our results support previous evidence that the Cx37 C1019T polymorphism is associated with the occurrence of MI. In addition to genetic risk factors, environmental influences are another likely explanation for between-study heterogeneity. The significant heterogeneity in the overall and subgroup analyses implies that there may be some clinical
Connexin37 Gene C1019T Polymorphism and Risk of Coronary Artery Disease
characteristics such as age, gender, lifestyle or other related diseases that need to occur before there is any effect or a substantial effect on effect size. We did identify that the proportion of DM had a significant effect on the effect size and the meta regression analysis found a tendency for increased CAD risk in populations with a higher incidence of DM. Increased glucose levels, as happen with diabetes, strengthen monocyte adhesion to endothelial cells via increased adhesion molecules and accelerate atherosclerosis (36). Although meta-regression suggests an ecological correlation rather than a causal inference, our results suggest that the Cx37 C1019T polymorphism and CAD risk are potentially correlated with DM. Further genetic association studies examining the influence of DM are required to accurately quantify the effect size. Despite the fact that our study was partially based on a relatively large sample size and conformed to HWE (37), there were some technical limitations (38). Possible publication bias could not be completely ruled out although the Egger’s test and the funnel plot did not indicate bias. Small negative results tend to be rejected for publication and the ‘‘grey’’ literatures (articles in languages other than English and Chinese) are likely not included. Funnel plot probes whether small studies with little precision give different results than larger studies with greater precision. Asymmetry in the funnel plot may occur as a result of an essential difference between the small and larger studies that arises from inherent between-study heterogeneity (39). We could not completely exclude the probability that small negative or unpublished studies were missing from the plot. In addition, we focused on a single nucleotide polymorphism and did not consider the possibility of gene-gene or haplotype-based effects (40,41). It is likely that the C1019T polymorphism moderately contributes to the risk of CAD patients and additional research is required to investigate whether this polymorphism, combined with other risk factors, will enhance the risk. In conclusion, the combined results of independent association studies suggest that the Cx37 C1019T polymorphism may be a modest risk factor for CAD, especially in MI patients. Moderate to large heterogeneity was identified between the studies and the proportion of DM was a statistically significant source for the observed variation. Our meta-analysis emphasized the necessity of taking genegene and gene-environment interactions into account when trying to interpret and combine data. More population surveys and function researches may facilitate understanding of the true relationship. Acknowledgments This work was supported by the Science and Technology Fund of Shanghai Jiao Tong University School of Medicine (11XJ21001) and the National Natural Science Foundation of China (81201839 and 81370401).
29
References 1. Gensini GF, Comeglio M, Colella A. Classical risk factors and emerging elements in the risk profile for coronary artery disease. Eur Heart J 1998;19(suppl A):A53eA61. 2. Topol EJ, Smith J, Plow EF, et al. Genetic susceptibility to myocardial infarction and coronary artery disease. Hum Mol Genet 2006;15(Spec No 2):R117eR123. 3. Crea F, Liuzzo G. Pathogenesis of acute coronary syndromes. J Am Coll Cardiol 2013;61:1e11. 4. Boerma M, Forsberg L, Van Zeijl L, et al. A genetic polymorphism in connexin 37 as a prognostic marker for atherosclerotic plaque development. J Intern Med 1999;246:211e218. 5. Koval M. Pathways and control of connexin oligomerization. Trends Cell Biol 2006;16:159e166. 6. Laird DW. Life cycle of connexins in health and disease. Biochem J 2006;394:527e543. 7. Saez JC, Berthoud VM, Branes MC, et al. Plasma membrane channels formed by connexins: their regulation and functions. Physiol Rev 2003;83:1359e1400. 8. Saez JC, Berthoud VM, Moreno AP, et al. Gap junctions. Multiplicity of controls in differentiated and undifferentiated cells and possible functional implications. Adv Second Messenger Phosphoprotein Res 1993;27:163e198. 9. Kwak BR, Mulhaupt F, Veillard N, et al. Altered pattern of vascular connexin expression in atherosclerotic plaques. Arterioscler Thromb Vasc Biol 2002;22:225e230. 10. Wong CW, Christen T, Kwak BR. Connexins in leukocytes: shuttling messages? Cardiovasc Res 2004;62:357e367. 11. Wong CW, Christen T, Roth I, et al. Connexin37 protects against atherosclerosis by regulating monocyte adhesion. Nat Med 2006;12:950e954. 12. Chanson M, Kwak BR. Connexin37: a potential modifier gene of inflammatory disease. J Mol Med (Berl) 2007;85:787e795. 13. Wang Q, Rao S, Shen GQ, et al. Premature myocardial infarction novel susceptibility locus on chromosome 1P34-36 identified by genomewide linkage analysis. Am J Hum Genet 2004;74:262e271. 14. Yamada Y, Izawa H, Ichihara S, et al. Prediction of the risk of myocardial infarction from polymorphisms in candidate genes. N Engl J Med 2002;347:1916e1923. 15. Lanfear DE, Jones PG, Marsh S,et al. Connexin37 (GJA4) genotype predicts survival after an acute coronary syndrome. Am Heart J 2007;154:561e566. 16. Egger M, Davey Smith G, Schneider M, et al. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997;315:629e634. 17. Anderson JL, Adams CD, Antman EM, et al. 2011 ACCF/AHA Focused Update Incorporated Into the ACC/AHA 2007 Guidelines for the Management of Patients With Unstable Angina/Non-STElevation Myocardial Infarction: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. Circulation 2011;123:e426ee579. 18. Joensen AM, Jensen MK, Overvad K, et al. Predictive values of acute coronary syndrome discharge diagnoses differed in the Danish National Patient Registry. J Clin Epidemiol 2009;62:188e194. 19. Wright RS, Anderson JL, Adams CD, et al. 2011 ACCF/AHA focused update incorporated into the ACC/AHA 2007 Guidelines for the Management of Patients with Unstable Angina/Non-ST-Elevation Myocardial Infarction: a report of the American College of Cardiology Foundation/ American Heart Association Task Force on Practice Guidelines developed in collaboration with the American Academy of Family Physicians, Society for Cardiovascular Angiography and Interventions, and the Society of Thoracic Surgeons. J Am Coll Cardiol 2011;57:e215ee367. 20. Cohn LD, Becker BJ. How meta-analysis increases statistical power. Psychol Methods 2003;8:243e253. 21. Lau J, Ioannidis JP, Schmid CH. Quantitative synthesis in systematic reviews. Ann Intern Med 1997;127:820e826. 22. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21:1539e1558.
30
Wu et al./ Archives of Medical Research 45 (2014) 21e30
23. Higgins JP, Thompson SG, Deeks JJ, et al. Measuring inconsistency in meta-analyses. BMJ 2003;327:557e560. 24. Biros E, Norman PE, Jones GT, et al. Meta-analysis of the association between single nucleotide polymorphisms in TGF-beta receptor genes and abdominal aortic aneurysm. Atherosclerosis 2011;219:218e223. 25. Yamada Y, Ichihara S, Izawa H, et al. Genetic risk for coronary artery disease in individuals with or without type 2 diabetes. Mol Genet Metab 2004;81:282e290. 26. Hirashiki A, Yamada Y, Murase Y, et al. Association of gene polymorphisms with coronary artery disease in low- or high-risk subjects defined by conventional risk factors. J Am Coll Cardiol 2003;42:1429e1437. 27. Han Y, Xi S, Zhang X, et al. Association of connexin 37 gene polymorphisms with risk of coronary artery disease in northern Han Chinese. Cardiology 2008;110:260e265. 28. Wong CW, Christen T, Pfenniger A, et al. Do allelic variants of the connexin37 1019 gene polymorphism differentially predict for coronary artery disease and myocardial infarction? Atherosclerosis 2007; 191:355e361. 29. Yeh HI, Chou Y, Liu HF, et al. Connexin37 gene polymorphism and coronary artery disease in Taiwan. Int J Cardiol 2001;81:251e255. 30. Hubacek JA, Stanek V, Gebauerova M, et al. Lack of an association between connexin-37, stromelysin-1, plasminogen activator-inhibitor type 1 and lymphotoxin-alpha genes and acute coronary syndrome in Czech Caucasians. Exp Clin Cardiol 2010;15:e52ee56. 31. Listi F, Candore G, Lio D, et al. Association between C1019T polymorphism of connexin37 and acute myocardial infarction: a study in patients from Sicily. Int J Cardiol 2005;102:269e271. 32. Morgan TM, Krumholz HM, Lifton RP, et al. Nonvalidation of reported genetic risk factors for acute coronary syndrome in a largescale replication study. JAMA 2007;297:1551e1561.
33. Seifi M, Fallah S, Ghasemi A, et al. Mutations of the connexin 37 and 40 gap-junction genes in patients with acute myocardial infarction. Clin Lab 2013;59:343e348. 34. Deo R, Nalls MA, Avery CL, et al. Common genetic variation near the connexin-43 gene is associated with resting heart rate in African Americans: a genome-wide association study of 13,372 participants. Heart Rhythm 2013;10:401e408. 35. Hemminki K, Lorenzo Bermejo J, Forsti A. The balance between heritable and environmental aetiology of human disease. Nat Rev Genet 2006;7:958e965. 36. Piga R, Naito Y, Kokura S, et al. Short-term high glucose exposure induces monocyte-endothelial cells adhesion and transmigration by increasing VCAM-1 and MCP-1 expression in human aortic endothelial cells. Atherosclerosis 2007;193:328e334. 37. Munafo MR, Flint J. Meta-analysis of genetic association studies. Trends Genet 2004;20:439e444. 38. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000;283:2008e2012. 39. Lau J, Ioannidis JP, Terrin N, et al. The case of the misleading funnel plot. BMJ 2006;333:597e600. 40. Wu Z, Lou Y, Jin W, et al. The C161T polymorphism in the peroxisome proliferator-activated receptor gamma gene (PPARgamma) is associated with risk of coronary artery disease: a meta-analysis. Mol Biol Rep 2013;40:3101e3112. 41. Wu Z, Lou Y, Jin W, et al. The Pro12Ala polymorphism in the peroxisome proliferator-activated receptor gamma-2 gene (PPARgamma2) is associated with increased risk of coronary artery disease: a metaanalysis. PLoS One 2012;7:e53105.