Association of Interferon Gamma +874T/A Polymorphism and Leukemia Risk

Abstract Interferon gamma (IFN-γ) has antitumor and antiproliferative effects, and previous studies indicated IFN-γ +874T/A (rs2430561) polymorphism were related to the risk of many types of cancer. However, the association between IFN-γ +874T/A polymorphism and leukemia risk remained controversial. We performed a comprehensive meta-analysis based on the Preferred Reporting Items for Systematic Reviews and Meta-analyses statement (PRISMA). Electronic database of Embase, Pubmed, and the Cochrane Library were searched for eligible articles published up to December 13, 2015. The association between genetic polymorphisms and leukemia risk was measured by odds ratios (ORs) and its corresponding 95% confidence intervals (CIs). A total of 8 studies amounting to 420 patients and 767 control subjects were retrieved for this study. Although associations between IFN-γ +874T/A polymorphism and overall leukemia risks were lacking, decreased chronic lymphocytic leukemia (CLL) risk was detected in the allelic model (T vs A, OR=0.660, 95%CI = 0.483–0.902, P = 0.009, I2 = 0.0% and P = 0.863 for heterogeneity), the codominant model (TT vs AA, OR = 0.472, 95%CI = 0.247–0.902, P = 0.023, I2 = 0.0% and P = 0.994 for heterogeneity), and dominant model (TT + TA vs AA, OR = 0.457, 95%CI = 0.285–0.734, P = 0.001, I2 = 40.3% and P = 0.195 for heterogeneity) by using fixed-effect model separately. On the contrary, results indicated T carries have an increased chronic myelogenous leukemia (CML) risk in dominant model (TT + TA vs AA, OR = 1.783, 95%CI = 1.236–2.573, P = 0.002, I2 = 19.0% and P = 0.295 for heterogeneity). This study suggests IFN-γ +874T/A polymorphism are related to CML and CLL risk. In addition, our work also points out IFN-γ +874T/A polymorphism may play dual contrasting role in leukemia risk.


INTRODUCTION
L eukemia is one of the most common malignancy in human, with an estimated 368,800 new cases and 303,200 deaths occurred in 2012 worldwide. 1 Moreover, leukemia is the leading cause of cancer death in those aged younger than 40 years among males and in children and adolescents (aged younger than 20 years) among females. 2 It is generally considered that both environmental and genetic risk factors contribute to leukemic progress; however, the biological mechanisms of leukemia are complex and have not been fully clarified. Cytokines and their genes play a fundamental role in the immunopathogenesis of malignant, infectious, and autoimmune disease. [3][4][5] It is, therefore, important to identify these genes variants contributing to leukemia pathogenesis.
Interferon gamma (IFN-g) is a T-helper 1 (Th1) proinflammatory cytokine that plays a pivotal role in antiviral activities and has antitumor and antiproliferative effects. 6 The production of IFN-g is under genetic control. The SNP (þ874T/A, rs2430561) of IFN-g gene, a T-to-A transition at þ874 loci in the 1st intron of IFN-g gene, is complete linkage disequilibrium with the12 CA repeat microsatellite and correlated with IFN-g production. 7 Moreover, the transcription factor nuclear factor kappa-B (NF-kB) binds preferentially to the þ874T allele, which is known to upregulate the expression of IFN-g. 8 The þ874T/A polymorphism generates 3 genotypes (TT, TA, and AA), and TT genotype produces a high level of IFN-g, helps the host's defense against viral infection. Conversely, the genotypes AA and AT cause relative low IFN-g production, which may increase the risk of viral infection. 7,8 It appears that the þ874T/A polymorphism can influence IFN-g expression, and in this way, affects immune response.
The relationship between IFN-g genes and leukemia was previously studied, 9 in addition, a number of studies have attempted to assess associations between IFN-g þ874T/A polymorphism and leukemia risk. [10][11][12][13][14][15][16] However, a unified conclusion is still unreached, furthermore, most of these studies had insufficient sample sizes of studied populations, which led to lower statistical power. Therefore, we conducted a metaanalysis to evaluate broadly the available evidence on the association between IFN-g þ874 T/A polymorphism and leukemia risk.

MATERIALS AND METHODS
This meta-analysis was conducted according to Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement (S1_Table, http://links.lww.com/MD/ A844), including the search strategy, selection criteria, data extraction, and data analysis. 17

Identification of Eligible Studies
Two investigators (YS and ZW) searched independently for all publications in Embase, PubMed, and Cochrane Library up to December 14, 2015, by using the following search terms: ''interferon gamma,'' ''IFN-g,'' ''rs2430561,'' ''polymorphism,'' and ''leukemia.'' Articles in reference lists were also hand-searched. Only human studies were searched and there was no restriction on the language.

Inclusion and Exclusion Criteria
The 2 investigators choose the suitable studies according to the following criteria: (1) Case-control or cohort design studies evaluated the associations between IFN-gþ874T/A polymorphism and leukemia risk, and the controls should be healthy populations.
(3) The articles did not include genotype frequencies, and this information was not available from the authors contacted. (4) Duplicate data.

Data Extraction
The based characters and data included the first author's name, publication date, country, ethnicity, diagnosis criteria for leukemia, source of control, total sample size, genotyping method, and the genotype frequencies of cases and controls were extracted independently by the 2 investigators (YS and ZW) from the included studies. To determine the accuracy of the extracted information, the data extracted by the 2 investigators should be all the same, and if there was a dispute they should check their data again. If the 2 investigators could not reach an agreement, the dispute was submitted to a 3rd reviewer (CL) to decide. In addition, if the genotype frequencies of cases and controls could not be extracted directly from the article, the 2 investigators should try to contact the authors for help, or obtain the data from previous literatures. HWE was calculated according to the genotype frequencies of controls. The quality of the studies included in this meta-analysis was assessed according to HWE, and we considered a high quality study should be fall in HWE.

Statistical Analysis
The association between genetic polymorphisms and leukemia risk was measured by ORs and its corresponding 95% CIs. The allelic model (T vs A), codominant model (TT vs AA, TA vs AA), dominant model (TT þ TA vs AA), and recessive model (TT vs TA þ AA) were examined.
Heterogeneity was assessed by a Chi-squared Q-test and I 2 statistics. 18,19 The fixed-effects model (the Mantel-Haenszel method) is used when the effects are assumed to be homogenous (P Q 0.1 or I 2 < 50%). Otherwise, the random effects model (the DerSimonian and Laird method) is used when they are heterogeneous (P Q < 0.1 or I 2 50%). Subgroup analysis was carried out stratified by leukemia type, ethnicity (Caucasian and Asian), and HWE (fall in HWE). Begg's funnel plot and Egger's test were used to investigate the publication bias in the metaanalysis; P < 0.05 indicated that the result was statistically significant. Sensitivity analysis was performed by omitting 1 study at a time.
All the tests in this meta-analysis were conducted with RevMan 5.2.7 (Cochrane Collaboration) and STATA software (version 12.0; Stata Corporation, College Station, TX).
An ethical approval was not necessary needed as this is a meta-analysis based on previous studies.

Study Characteristics
The literature search and study selection procedures are shown in Figure 1. Leukemia types, genotype numbers, and HWE examination results are shown in Table 1. According to the inclusion and exclusion criteria, 7 articles were included in the meta-analysis. The article performed by Baştürk et al 11 was separated as 2 studies for they evaluated 2 different diseases, therefore, a total of 8 studies amounting to 420 patients and 767 control subjects were retrieved for this study. Two studies populations come from Asian 10,12 and 6 studies were Caucasian. 11,[13][14][15][16] Two studies were failed to fall in HWE. 10,14 Meta-Analysis for the Association Between the IFN-g R874A/T Polymorphism and Leukemia The meta-analysis results for all the studies are shown in Table 2. When all the eligible studies were pooled into the metaanalysis of IFN-g þ874A/T polymorphism, there was no significant association between IFN-g þ874A/T polymorphism and leukemia risk under all comparison models.

Evaluation of Heterogeneity and Publication Bias
Significant heterogeneity was discovered in this study expect the recessive model in the overall population. Then we carried out subgroup analysis according to leukemia type, and the results showed there was no obvious heterogeneity in the subgroup analyses of CLL and CML, expect TT vs TA analysis (I 2 > 50%, Table 2), suggesting that leukemia type was the major source of heterogeneity in our meta-analysis. Figure 3 showed the Begg's funnel plot in allelic model. No significant publication bias was detected using Begg funnel plot in the overall population in all comparison models. The statistical results of Egger's test also proved the evidence of funnel plot symmetry.

Sensitivity Analysis
Sensitivity analysis was performed by omitting 1 study at a time, and the estimate of overall effect did not change appreciably. In addition, when we excluded the studies out of HWE, the statistical significance of the results did not change (Table 2).

DISCUSSION
This is the 1st quantitative evaluation on IFN-g þ874T/A polymorphism and leukemia risk. Results based on the 420 leukemia cases and 767 matched controls from 8 studies suggested that IFN-g þ874T/A polymorphism were related to CML and CLL susceptibility. In this study, although associations between IFN-g þ874T/A polymorphism and overall leukemia risks were lacking, higher CLL susceptibility was detected in allelic model, codominant model, and dominant model, and results suggested T carries (patients with TT or TA genotype) have a decreased risk of CLL compared with AA genotype. On the contrary, results from dominant model indicated T carries have nearly 1.783-fold increased CML risk. To our best knowledge, this is the 1st evidence suggested IFN-g þ874T/A T allele increased cancer risk from meta-analysis.
T allele in the þ874 position of IFN-g is a protect genotype for CLL is biologically plausible. First, the þ874 position, which is in strong linkage disequilibrium with the 12 CA repeat microsatellite in the noncoding region of the 1st intron of IFN-g gene, controls the production of IFN-g. 20 As we known now, IFN-g is a proinflammatory cytokine playing a pivotal role in both innate and adaptive immune responses, and has antitumor and antiproliferative effects. 6 Therefore, it is understandable that T allele, which with high level IFN-g production, may be a protect gene for CLL. Second, the DNA sequence containing the þ874 T allele is the specific binding site for NF-kB. 21 NF-kB knockout mice have been shown to have low values of IFNg and an increased susceptibility to infection. 22 This polymorphism leading to the dysfunction of NF-kB pathway may cause oxidative damage, which also can increase the risk of cancer. 21 Indeed, our finding is consistent with previous metaanalysis that T allele is a protect genotype among patients in breast cancer, cervical cancer, asthma, and leprosy. [23][24][25][26] However, this study observed T allele is a risk genotype for CML. These results suggested 2 faces of IFN-g in leukemia. Indeed, T allele is a risk genotype for diseases had been observed in bipolar disorder, 27 and this is also not a new concept for the roles of cytokines in cancer. Transforming growth factor (TGF)-b has an ability to induce malignant behavior of normal fibroblasts, at the same time, it has profound growthsuppressive effects on many cells. 28 Tumor necrosis factor (TNF), an extraordinarily pleiotropic cytokine, also displays pro-and antitumoral effects. 29 A potential mechanism is localized inflammation may lead to an immunosuppressive and tolerogenic tumor microenvironment. 30 For the dark side of IFN-g, Zaidi and Merlino 31 believed it may lie in the homeostatic functions of IFN-g. Although high level of IFN-g can protect normal cells from the collateral damage, these same mechanisms may also allow cancer cells to evade attack from the immune system. 31 NK cells play a central role in tissue repair, however, Taniguchi et al 32 found IFN-g can enhanced resistance to NK cells, which was confirmed by another study and possible mechanism is higher metastatic ability of IFN-g clones attributed to increased resistance to NK cells. 33 But evidence so far for the roles of IFN-g that may be more protumorigenic has been indirect, the exact mechanisms might IFN-g behave as a ''bad guy'' in cancer need to be confirmed by further studies.
Genetic susceptibility is believed to be an important factor in the development of cancers. 34 Meanwhile, the predisposition of every cancer types could be differentially influenced by genetic factors. Previous studies have showed the association between IFN-g þ874T/A polymorphism and cervical cancer, 35,36 bladder cancer, 37 liver cancer, 38 and breast cancer, 39 and suggested that T allele might be a protective genotype in cancer for TT genotype produces a high level of IFN-g. Yet a recent meta-analysis showed no association between IFN-g þ874T/A polymorphism and cancer susceptibility, 40 at the same time, the other meta-analysis found positive association in cervical cancer and breast cancer, 23,26 which suggested that IFN-g þ874T/A polymorphism play different roles in different cancer types; therefore, the conclusion from a pooled analysis for association between IFN-g þ874T/A polymorphism and all cancer types may be unreliable. This metaanalysis confirmed this point. We found no significant association between IFN-g þ874A/T polymorphism and leukemia risk when all leukemia types were pooled into the metaanalysis, however, when stratified by leukemia types, results suggested T allele contributed completely opposite influence between CML and CLL. These results not only could partly explain the inconsistent results in different leukemia types but are also the reason why we conducted this study.
This study performed a subgroup analysis stratified by ethnicities and the significant associations were only found in the dominant model among Asians. Therefore, the genetic background may be another important factor contributed to the inconsistent results. Indeed, genetic background is distinct among different populations and our previous study had shown the genotype and allele frequencies of IFN-g þ874T/A were quite different among different regions and ethnicities. 4 However, we noticed that there are 2 studies on CLL and 3 on CML in Caucasians, which may lead to negative results happened in the pooled analysis considering the dual role of IFN-g þ874T/A in CLL and CML.
Genome-wide association studies (GWAS) have successfully identified 13 loci associated with risk of CLL/small lymphocytic lymphoma before 2011. 41   Our results suggested that rs2430561 in IFN-g on chromosome 12q24 contributes to susceptibility to CML and CLL based on traditional genotyping methods, and these results have motivated us to carry out a GWAS focus on the SNPs on chromosome 12 to discover novel susceptibility loci for leukemia in the future.
In clinical practice, patients with leukemia are much less than that of lung cancer, liver cancer, breast cancer, and cervical cancer; therefore, the most limitation of this meta-analysis is that the present studies on susceptibility of leukemia are all with relative small sample sizes. As an example, we noticed the TA genotype frequency is only 1 in the study by Amirzargar et al, 10 which can easily lead to false positive or negative results. Thus, we need to be taken into consideration when interpreting the findings. Further study with larger sample numbers will be helpful to improve the power for detecting the positive effects. In addition, other confounders, such as age and gender, were not adjusted because it is hard to get original data from author. In spite of these, our meta-analysis had some advantages. In the first place, this study greatly increased the statistical power compared with individual studies. There is one more point, no significant heterogeneity was detected in subgroup analysis stratified by leukemia types, together with publication bias and sensitivity analysis results indicated that our results were moderately robust.

CONCLUSIONS
Taken together, results from this meta-analysis demonstrate that IFN-g þ874T/A polymorphism contribute to CML and CLL risk. However, our results are still needed to strengthen by further studies with larger sample sizes.