A single nucleotide polymorphism in the interferon-γ gene (IFNG +874 T/A) is associated with susceptibility to tuberculosis

Interferon-γ (Interferon gamma, IFNG) is an important cytokine involved in providing resistance to mycobacterial diseases. Common variants of IFNG, such as IFNG +874 T/A (rs2430561), may be related to tuberculosis susceptibility, but this association has not been consistently observed. We performed an updated metaanalysis to evaluate the association between the IFNG +874 T/A (rs2430561) polymorphism and tuberculosis susceptibility. PubMed and SinoMed databases were searched up to October 2016, and odds ratios (OR) and 95% confidence intervals (CI) were used to assess the association strength. Based on search criteria for manuscripts reporting tuberculosis susceptibility and its relationship with the IFNG +874 T/A (rs2430561) polymorphism, 42 case-control studies from 39 different articles were retrieved. Significantly positive, decreased, and protective associations were found between the IFNG +874 T/A (rs2430561) polymorphism and tuberculosis risk in five genetic models. Moreover, in the stratified subgroup analysis, a protective relationship was detected in four different ethnicities and sources of the control groups. Furthermore, the IFNG +874 T/A (rs2430561) polymorphism played an important role in protecting individuals from both pulmonary tuberculosis and extra-pulmonary tuberculosis. Our meta-analysis suggests that the IFNG +874 T/A (rs2430561) polymorphism is potentially associated with tuberculosis susceptibility and may be used as a predictive biomarker. Further studies with larger sample sizes and consideration of gene-environment interactions should be conducted to elucidate the role of IFNG +874 T/A (rs2430561) polymorphism in tuberculosis susceptibility.


INTRODUCTION
Tuberculosis (TB) is a chronic infectious disease caused by Mycobacterium tuberculosis (MTB). Data from the global tuberculosis report estimated a worldwide incidence of 8.7 million cases, most of which occurred in Asia and Africa [1]. One-third of the world's population is potentially infected with M. tuberculosis, but only 10% of the infected individuals eventually develop clinical TB, indicating that the host's genetic and other factors (such as alcohol consumption, malnutrition, and human immunodeficiency virus infection) may play essential and complex roles in determining susceptibility and progression to tuberculosis [2][3][4]. An increasing number of studies have focused on the relationship between genetic variations of several genes and TB susceptibility [5,6].
Currently, the gene encoding interferon-γ (IFNG) is the most extensively studied gene regarding tuberculosis susceptibility. Human IFNG is located on chromosome 12 (12q14) and has four exons spanning approximately 6

Research Paper: Immunology
Oncotarget 50416 www.impactjournals.com/oncotarget kb. Interferon-γ (IFN-γ), which is encoded by IFNG and produced by T helper 1 (Th1) lymphocytes, is upregulated and secreted as a major cytokine to activate macrophages, and is critical for the control of M. tuberculosis infection [7]. Cases of active TB are characterized by decreased production of IFN-γ from the peripheral blood mononuclear cells compared with that observed in latent infection. In addition, local and systemic IFN-γ levels correlate with the severity of the disease [8].
Lio et al. first noted a significant association between the +874A/T polymorphism (rs2430561) in IFNG and protection against TB in Sicilia (P < 0.05) [9]. The SNP +874 (A/T) is located at the 5′-end of a CA repeat in the first intron of the human IFNG [10]. The +874 T-allele is linked to the 12 CA repeats, whereas the A-allele is linked to the non-twelve CA repeats [11]. The specific sequence of the T-allele provides a binding site for the transcription factor nuclear factor-κB [12,13]. As nuclear factor-κB induces IFN-γ expression, the presence of the T-allele correlates with high IFN-γ expression, whereas that of the A-allele correlates with low expression.
Many epidemiologic studies, including metaanalyses [14,15], suggested associations between IFNG +874 T/A (rs2430561) polymorphism and the risk of developing TB. However, the data were inconsistent, partly because of differences in study populations and case ascertainment criteria, or small sample sizes with a high rate of false-positives and limited power of detecting modest associations. Considering the important role of IFNG +874 T/A (rs2430561) polymorphism in TB development, we studied all eligible case-control studies that included characteristics such as ethnicity of the patients, types of tuberculosis, and control sources.

Study characteristics
Using various combinations of key terms, 152 article titles were retrieved by a literature search using the PubMed and SinoMed databases. As shown in Figure  1, 66 articles were excluded after screening the abstracts of the manuscripts. The full texts were evaluated, and 47 additional articles were excluded because of the following   Oncotarget 50418 www.impactjournals.com/oncotarget reasons: duplication (two), meta-analysis (seven), review (twelve), discussion on polymorphism in other diseases (tegumentary leishmaniasis and leprosy) (two), other SNPs in IFNG (one), IFNG receptor gene polymorphisms (fifteen), and other genes (three), and lacked case-control study (five). Finally, 39 different articles [9, concerning IFNG +874 T/A (rs2430561) polymorphism and TB susceptibility were included in our meta-analysis (Table 1). Overall, 42 case-control studies with 8,574 cases and 9,011 controls were retrieved based on the search criteria. The controls were mainly healthy populations.   [20,25,30,34,40,42,47], the distribution of genotypes in the controls in remaining 35 studies was in agreement with HWE, and it reports the meta-analysis results of studies showing no deviation from the HWE.
Oncotarget 50419 www.impactjournals.com/oncotarget 0.382 in the case and control group, respectively, both lower than the results in the 1000 Genomes Browser database. Then, the frequency of the T-allele was found to be significantly higher in case individuals of Caucasian ethnicity than in those of African or mixed ethnicities (41.3% vs. 22.9%, P = 0.005, or 41.3% vs. 27.2%, P = 0.017, respectively). A similar trend was found in the control group (Figure 2). Except for seven studies [16,21,26,30,36,38,43], the distribution of genotypes in all the controls was in agreement with Hardy-Weinberg equilibrium (HWE).

Quantitative synthesis
In the overall analysis, a significant protective association could be observed between TB risk and the variant genotype of IFNG +874 T/A (rs2430561) polymorphism in different genetic models. For example, for the additive model (TA vs. AA), odds ratios (OR) = 0.71, 95% confidence intervals (CI) = 0.63-0.80, P-value of heterogeneity test (P h ) = 1×10 -6 , and in the dominant model, OR = 0.68, 95% CI = 0.60-0.77, P h = 1×10 -6 ( Figure 3, Table 2). If some studies that do not conform to HWE were removed, these associations were not affected.
In the ethnicity type subgroup, this pattern of  Table 2).
To improve our study, finally we made stratified subgroup analyses according to genotyping methods and sample size. Some positive associations were also found in this section (Table 2).

Bias diagnosis and sensitivity analysis
The Begg's funnel plot and Egger's test were performed to access the publication bias of the literature. The shape of the funnel plot did not reveal any obvious asymmetry, and the Egger's test suggested the absence of publication bias. For example, in the additive model (TA vs. AA) analysis, z = 1.11, P = 0.269 for Begg's test and t = −0.87, P = 0.389 for Egger's test (Figures 7, 8, Table 3). We used a sensitivity analysis to determine whether modification of the meta-analysis inclusion criteria affected the results. No single study qualitatively influenced the summary ORs, as indicated by the sensitivity analysis (for example, the allelic effect, Figure  9).

DISCUSSION
To understand the pathogenesis of TB and better predict individual risk, it is important to identify the SNPs that affect the function of genes contributing to TB susceptibility. Genome-wide association studies (GWAS) allow genotyping of frequent genetic polymorphisms in the genome without assuming the genomic location of the causal variants. As most of the genome has been surveyed, GWAS eliminates the disadvantages associated with the single polymorphism or candidate gene approach where only a few polymorphisms are investigated [54]. However, GWAS may identify false associations, and therefore, replication or validation of the results in independent populations is essential [55]. A Wellcome Trust Case-Control Consortium-based study initiated GWAS that considered up to 2,000 cases and controls each for tuberculosis and malaria. This study identified NRAMP1, IFNG, NOS2A, MBL, VDR, and certain TLR-associated genes as important for susceptibility to TB and malaria, and the list has been growing ever since [55][56][57].
Lio et al. [9] were the first to investigate the association between IFNG +874 T/A (rs2430561) polymorphism and the risk of developing TB. Since then, other researchers have reproduced their work on IFNG in different populations and for two disparate types of TB. However, the results were confounding, even within populations. Meta-analysis provides a means for effectively increasing the size of the sample Oncotarget 50425 www.impactjournals.com/oncotarget by pooling data from individual correlation studies, thus enhancing the statistical power of the analysis to estimate genetic effects [58]. We used this method to demonstrate statistically significant genetic associations.
To the best of our knowledge, this is the first systematic meta-analysis exploring the associations between IFNG polymorphism and TB risk. Our analysis involved about 8,574 individuals with TB and 9,011 controls. The main discovery of our study was that the T-allele of the IFNG +874 T/A (rs2430561) polymorphism showed a protective effect for TB susceptibility, which was consistent with the observation that the presence of the T-allele correlates with high IFN-γ expression and increased resistance to M. tuberculosis infection, whereas the A-allele correlates with low expression. The highlight of our article is that we classified TB into PTB and EPTB, which assisted in the analysis of effective target sites for different types of TB, and thereby, we provided a novel concept compared to the previously published metaanalyses [14,15].
Although considerable efforts and resources have been invested in examining the possible associations between IFN gene polymorphisms and TB risk, certain inherent limitations exist. First, although we collected all eligible studies, the sample sizes of these studies were not large enough to provide adequate statistical power to evaluate the association between this polymorphism and TB risk, particularly for certain ethnicities. Thus, not only was there an increased likelihood of type I/ II error, but also a lack of sufficient statistical power to evaluate the association between these polymorphisms and tuberculosis risk. Second, because of the complex nature of the immune system and the polygenic nature of complex diseases, such as TB, it has become increasingly evident that gene-gene interactions play a far more important part in an individual's susceptibility to tuberculosis than single polymorphisms would on their own [59]. Specific environmental and lifestyle factors, including age, sex, smoking habits, familial history, and disease stage, may possibly alter the associations between IFNG polymorphisms and TB. Therefore, it is necessary to evaluate the roles of specific environmental factors and lifestyles in disease prognosis.
In spite of these drawbacks, our meta-analysis provided several advantages. First, a substantial number of cases and controls were pooled from different studies, which significantly increased the statistical power of the analysis. Second, no publication bias was found and the Newcastle-Ottawa Scale (NOS) scores of the included studies indicated high quality. Third, our study contained the highest number of individuals so far, which conferred increased statistical power and credibility. The power of this meta-analysis of +874 T/A (rs2430561) polymorphism was greater than 0.95, which suggested that the data was trustworthy.
In conclusion, our present meta-analysis suggests that the +874 T/A (rs2430561) polymorphism in IFNG may be associated with the risk of developing TB. Further studies that consider additional gene-environment interactions and are based on larger sample sizes should be conducted to elucidate the role of IFNG polymorphisms in the etiology and clinical characteristics of TB development.

Search strategy
We searched the PubMed and SinoMed (http://www. sinomed.ac.cn/) databases through October 2016, using the keywords "tuberculosis," "TB," "polymorphism," and "IFN-gamma." One hundred and fifty-two articles were retrieved, 39 of which satisfied our inclusion criteria. We also screened references of the retrieved articles and reviewed all articles via a manual search.

Inclusion and exclusion criteria
For inclusion, studies were required to (i) assess the association between TB risk and IFNG +874 T/A (rs2430561) polymorphism; (ii) be case-controlled, and (iii) contain all three kinds of genotypes (TT, TA, and AA) for cases and controls. Studies were excluded if they (i) included no control population; (ii) did not report genotype frequency data, and (iii) were duplicated publications.

Quality score assessment
The NOS [60] was selected to assess the quality of each study. This measure assesses aspects of the methodologies used in observational studies, which are related to the study quality, including selection of cases, comparability of populations, and ascertainment of exposure to risks. The NOS rating ranges from zero stars (worst) to nine stars (best). Studies with a score of seven stars or greater were considered to be of high quality.

Data extraction
Two authors independently extracted the data that complied with the selection criteria. These data included the first author's last name, year of publication, country of origin, ethnicity of the study population, TB type, genotypes of both the case and control groups, source of control, HWE/ MAF in both case and control groups, and genotyping method. Ethnicity was categorized as Caucasian, Asian, African, and mixed population. The control subgroups were population-based (PB) and