Association of tumor necrosis factor-α gene polymorphisms and coronary artery disease susceptibility: a systematic review and meta-analysis

Background The goal of this study was to review relevant case-control studies to determine the association of tumor necrosis factor-α (TNF-α) gene polymorphisms and coronary artery disease (CAD) susceptibility. Methods Using appropriate keywords, we identified relevant studies using PubMed, Cochrane, Embase, CNKI, VANFUN, and VIP. Key pertinent sources in the literature were also reviewed, and all articles published through April 2019 were considered for inclusion. Based on eligible studies, we performed a meta-analysis of association between 308G/A, 238G/A, 857C/T, 863C/A and 1031 T/C polymorphisms in TNF-α and risk of CAD. Results We found 25 studies that were consistent with this meta-analysis, including 7697 patients in the CAD group and 9655 control patients. TNF-α 308G/A locus A showed no significant association with CAD susceptibility by the five models in the analysis of the overall population, European, African, South Asian, and North Asian patients. TNF-α 863C/A locus A and 1031 T/C locus C exhibited no significant association with CAD susceptibility. TNF-α 238G/A locus A had no significant association with CAD susceptibility in the overall population. However, TNF-α 238G/A locus A showed significant association with higher CAD susceptibility in the subgroup of Europeans and north Asians. TNF-α 857C/T locus T had no significant association with CAD susceptibility in the analysis of the overall population and Europeans. In the north Asian population, TNF-α 857C/T locus T was associated with lower CAD susceptibility by the heterozygote model. Conclusion TNF-α 308G/A, 857C/T, 863C/A, and 1031 T/C has no significant association with CAD susceptibility. TNF-α 238G/A locus A has significant association with CAD susceptibility in Europeans and north Asians, but has no significant association in the overall population. Studies with a larger sample size are required to confirm the association between TNF-α 238G/A and CAD susceptibility.


Background
Coronary artery disease (CAD) refers to a heart disease caused by ischemia and hypoxia of myocardial cells following coronary artery stenosis or blockage due to coronary atherosclerosis (AS). Globally, CAD is an important cause of mortality and morbidity, with approximately 9 million deaths between 2007 and 2017 [1]. At present, the major risk factors for CAD confirmed in clinical studies include age, gender, poor diet and lifestyle habits, metabolic syndrome (including obesity or overweight, hypertension, type 1 or type 2 diabetes and dyslipidemia), smoking, drinking, psychosocial factors and genetic factors. Studies [2,3] showed that the risk of developing CAD in an individual is modulated by an interplay between genetic and lifestyle factors. In the future, genetic testing can be expected to enable precision medicine approaches by identifying subgroups of patients at increased risk of CAD or those with a specific driving pathophysiology in whom a therapeutic or preventive approach is most useful.
Tumor necrosis factor (TNF) is a proinflammatory cytokine in vivo with extensive biological activities. Human TNF gene, located in the short arm of chromosome 6, is a 7 kb DNA sequence composed of TNFA and TNFB, encoding TNF-α and TNF-β, respectively, each containing 4 exons and 3 introns. At present, many scholars agree that there is an interactive feedback loop between acute or chronic inflammatory reactions, the dynamics of atherosclerotic plaques, platelet aggregation, activation of the coagulation system and lipid metabolism disorders. Inflammatory response may be an important trigger mechanism, and there are many kinds of inflammatory biomarkers in serum, including C-reactive protein, intercellular adhesion molecule, p-selectin, amyloid A protein, fibrinogen, e-selectin, pregnancyrelated plasma protein-a, serum interleukin-6, and TNF-α [4][5][6]. Studies have shown that the presence of TNF-α gene polymorphism may affect gene transcription and expression levels, and is associated with a variety of diseases such as rheumatoid arthritis, type 1 diabetes, type 2 diabetes, ankylosing spondylitis, sarcoidosis, and silicosis [7][8][9]. The aim of this study was to perform a meta-analysis of all available literature to obtain updated evidence about association between TNF-α polymorphisms and CAD susceptibility.

Search strategy
To identify studies pertaining to the associations between 308G/A, 238G/A, 857C/T, 863C/A and 1031 T/C polymorphisms in TNF-α and risk of CAD, we reviewed the Cochrane library, PubMed, Embase, CNKI, VANFUN, and VIP databases for relevant articles published through April 2019. We also reviewed the references of all identified articles to look for additional studies. Search terms were as follows: gene polymorphisms, gene, polymorphism, variant, genotype, tumor necrosis factor-α, TNF-α, coronary artery disease, CAD, angina, myocardial infarction, ischemic heart disease, tumor necrosis factor and TNF. These terms were used in combination with "AND" or "OR". This literature review was performed independently by two investigators, with a third resolving any disputes as needed. The detailed search strategy of PubMed: ("gene polymorphisms" or "gene" or "polymorphism" or "variant" or "genotype") and ("tumor necrosis factor-α" or "TNF-α" or "tumor necrosis factor" or "TNF") and ("coronary artery disease" or "CAD" or "angina" or "myocardial infarction" or "ischemic heart disease") AND Humans [Mesh]Search.
Following the PICOS (Participants, Interventions, Comparisons, Outcomes and Study design) principle, the key search terms included (P) patients with CAD; (I) detection the gene polymorphisms of TNF-α; (C/O) compare the gene polymorphisms of TNF-α between the CAD group and the control group; (S) case-control studies or cohort study.

Study selection criteria
Eligible studies met the following criteria: [1] casecontrol or cohort studies [2]; the subjects in the case group were patients with CAD [3]; the participants in the control group did not have CAD [4]; 308G/A, 238G/ A, 857C/T, 863C/A and 1031 T/C of TNF-α were studied; 4) English or Chinese language.  Studies were excluded for meeting the following criteria: [1] duplicate articles or results [2]; apparen tdata errors [3]; case reports, theoretical research, conference reports, systematic reviews, meta-analyses, and other forms of research or comment not designed in a randomized controlled manner [4]; irrelevant outcomes [5]; lack of a control group.
Two investigators independently determined whether studies met the inclusion criteria, with a third resolving any disputes as needed.

Data extraction and quality assessment
For each included study, two categories of information were extracted: basic information and primary clinical outcomes. Basic information relevant to this metaanalysis included: author names, year of publication, country, ethnicity, and sample size. Primary outcomes relevant to this analysis included frequency of genotypes (308G/A, 238G/A, 857C/T, 863C/A and1031T/C of TNF-α) in the CAD group and the control group. This data extraction was performed independently by two investigators, with a third resolving any disputes as needed.
We used Newcastle-Ottawa Scale (NOS) to assess the quality of eligible studies. The version of casecontrol studies included a set of questions: adequacy of case definition, representativeness of cases, selection of controls, definition of controls, matched age and sex, additional factors, ascertainment of exposure, case and controls (the same ascertainment method), cases and control (the same non-response rate).
Statistical analysis STATA v12.0 (TX, USA) was used for all analyses. Heterogeneity in study results was assessed using chi-squared and I 2 tests and appropriate analytic models (fixed-effects or random-effects) were determined. A chi-squared P ≤ 0.05 and an I 2 > 50% indicated high heterogeneity and the random-effects model was used in this case. A chi-squared P > 0.05 and an I 2 ≤ 50% indicated acceptable heterogeneity and the fixed-effects model was used. Egger's test and Begg's test were used to determine whether there was publication bias. Under ideal conditions (such as random mating, no selection, mutation, or migration), if the population is in line with the Hardy-Weinberg equilibrium (HWE), the proportion of certain characteristic genes will remain unchanged in inheritance. HWE is closely related to genotyping quality. HWE is a common hypothesis. In the metaanalysis of genetic association study, it is necessary to test whether the genotype distribution of the control group conforms to HWE. If the HWE genetic balance test was not provided in the original text or not performed on the control group, we used Stata v12.0 to carry out manual detection and extracted the corresponding results (P value). Five commonly used gene models were selected for meta-analysis: the allelic model (A vs. C); homozygote model (AA vs. CC); heterozygote model (AC vs. CC); dominant model (AA + ACvs.CC); regressive model (AA vs. AC+ CC). OR and 95% CI were used to analyze all the indexes.

Overview of included studies
We reviewed a total of 1115 articles identified by our initial keyword search, of which 1026 were excluded following title/abstract review. The complete full texts of the remaining 89 articles were assessed, excluding 64 articles that did not meet the study inclusion criteria. Reasons for exclusion of these studies were theoretical research [3], lack of clinical outcomes [10], duplicate articles [5], and lack of a control group [11]. We ultimately identified a total of 25 case-control studies  that met the inclusion criteria for this meta-analysis, including 7697 patients in the CAD group and 9655 in the control group. The study selection process is outlined in Fig. 1. Table 1 summarizes the basic information for each study, including author names, year of publication, country, ethnicity, and sample size. Seven studies involved Eurpeans, 14 involved north Asians, 3 involved south Asians, 2 involved Africans, and 1 involved North Americans. The risk of bias assessed by NOS is presented in Fig. 2.
Meta-analysis of TNF-α308G/a polymorphisms and CAD susceptibility In total, 19 studies with 7036 patients in the CAD group and 8940 controls reported on the association between TNF-α 308G/A and CAD susceptibility. For studies without significant heterogeneity (chi-squared P > 0.05  In the subgroup analysis, there was no significant association between TNF-α 308G/A locus A and CAD by the five models.
All the above results are presented in Fig. 3, Fig. 4 and Table 2.

Meta-analysis of TNF-α 238G/a polymorphisms and CAD susceptibility
In total, 12 studies with 5167 patients in the CAD group and 7103 controls reported on the association of TNF-α 238G/A and CAD susceptibility. For studies without significant heterogeneity (chi-squared P > 0.05 and I 2 < 50%), the fixed-effects model was chosen to analyze the all the comparison models except the dominant model and the heterozygote model in the subgroup analysis of the overall population, north Asians and HWE, and the allelic model in the subgroup analysis of HWE. All the above results are presented in Fig. 5 and Table 3.

Meta-analysis of TNF-α 857C/T polymorphisms and CAD susceptibility
In total, 9 studies with3843 patients in the CAD group and 5616 in the control group reported on the association of TNF-α 857C/T and CAD susceptibility. For studies with no significant heterogeneity (chi-squared test P > 0.05 and I 2 < 50%), the fixed-effects model was chosen to analyze all the comparison models. The results of Begg's test (p > 0.05) revealed no significant publication bias among the study results. The results showed no significant association between TNF-α 857C/T locus T and CAD susceptibility: All the above results are presented in Fig. 6 and Table 4.

Meta-analysis of TNF-α 863C/a polymorphisms and CAD susceptibility
In total, 10 studies with3225 patients in the CAD group and 4784 controls reported on the association of TNF-α 863C/A and CAD susceptibility. For studies All the above results are presented in Fig. 7 and Table 5.

Meta-analysis of TNF-α 1031 T/C polymorphisms and CAD susceptibility
In total, 9 studies with 3851 patients in the CAD group and 3936 controls reported on the association between TNF-α 1031 T/C and CAD susceptibility. For studies with no significant heterogeneity (chisquared test, P > 0.05 and I 2 < 50%), the fixed-effects model was chosen to analyze all the comparison model except the regressive model and homozygote model. The results of Begg's test (p > 0.05) showed no significant publication bias among the study results.
The results showed no significant association between TNF-α 1031 T/C locus C and CAD susceptibility: the al- All the above results are presented in Fig. 8 and Table 6.

Discussion
Atherosclerosis is the pathological basis of coronary heart disease, and inflammation plays a crucial role in the occurrence and development of atherosclerosis. Inflammation plays an important role in the formation, growth, rupture, and/or wear and tear of atherosclerotic plaques and the formation of blood clots. In particular, acute cardiovascular events such as heart failure, nausea and arrhythmia, cardiogenic shock and even cardiac arrest caused by plaque rupture and secondary acute thrombosis leading to complete occlusion of blood vessels are common clinical emergencies with sudden onset and high mortality. Therefore, the occurrence and development of coronary heart disease is a process of chronic inflammatory response. TNF-α is an important proinflammatory cytokine mediating inflammatory response and immune regulatory response in vivo. TNF-α can affect the development of coronary heart disease through the following ways: [1] participation in the inflammatory response of atherosclerotic plaques, the formation and rupture of plaques, leading to coronary heart disease and even acute myocardial infarction [2]. Direct injury to vascular endothelial cells can increase their permeability, and blood cholesterol can easily penetrate the intima and deposit in the wall of the vessels [3]. Promotion of proto-oncogene transcription, production of platelet-derived growth factors, disruption of the balance between blood coagulation and anti-blood coagulation, and promotion of thrombosis [4]. Inhibiting lipoprotein enzyme activity is not conducive to lipid dissolution and deposition in the vascular wall, promoting the formation of arteriosclerosis and aggravating the damage of the vascular wall. TNF-α polymorphic loci are located in the promoter region of − 308, − 238, − 163, − 244,-857, − 836, − 1031 and other loci. The presence of these gene polymorphisms may affect gene transcription and expression levels and be associated with various diseases.
In previous studies, Fengtian et al. [35] included 14 studies and found no association between T-1031C, C-857 T and C-863A and CAD risk. Karely et al. [36] included 27 articles, and found a significant association between TNF-a G308A and CHD in the whole population, and between the variant G238A and CHD in the Asian population. In our study, we found that TNF-α 308G/A locus A had no significant association with CAD susceptibility by the five models in the analysis of the overall population, Europeans, Africans, south Asians, and north Asians, which is contrary to the conclusion of Karely Pulido-Gomez. TNF-α 863C/A locus A and 1031 T/C locus C showed no significant association with CAD susceptibility, which is consistent with the conclusion of Fengtian HUANGFU. TNF-α 238G/A locus A had no significant association with CAD susceptibility in the overall population. However, TNF-α 238G/A locus A displayed significant association with higher CAD susceptibility in the subgroup of Europeans and north Asians. The association of TNF-α 238G/A in Asians is consistent with the study by Karely Pulido-Gomez. TNF-α 857C/T locus T had no significant association with CAD susceptibility in the analysis of the overall population and Europeans. In the north Asian  population, TNF-α 857C/T locus T was associated with lower CAD susceptibility. However, there are certain limitations to the present analysis, which are as follows: [1] only English and Chinese articles were included [2]; individual studies had different exclusion/inclusion criteria [3]; the severity of CAD was varied in different studies [4]; the number of included studies was limited, and some of the studies had a small sample size [5]; pooled data were analyzed, as individual patient data was not available, precluding more in-depth analyses.

Conclusion
Our results indicate that TNF-α 308G/A, 857C/T, 863C/ A, and 1031 T/C are not associated with CAD susceptibility. TNF-α 238G/A locus A has significant association with CAD susceptibility In Europeans and north Asians, but has no significant association in the overall population. In the north Asian population, TNF-α 857C/T locus T was associated with lower CAD susceptibility. Larger-sample studies are required to confirm the association between TNF-α 238G/A and 857C/T and CAD susceptibility.