The MPO−463G>A Polymorphism and Lung Cancer Risk: A Meta-Analysis Based on 22 Case–Control Studies

Background Myeloperoxidase (MPO) is an endogenous oxidant enzyme that produces reactive oxygen species (ROS) and may be involved in lung carcinogenesis. The MPO−463G>A polymorphism influences MPO transcription and has been associated with lung cancer susceptibility. However, the association between the MPO−463G>A polymorphism and lung cancer risk remains controversial. Method To investigate the effect of this polymorphism on lung cancer susceptibility, we performed a meta-analysis based on 22 published case–control studies including 7,520 patients with lung cancer and 8,600 controls. Odds ratios (ORs) with 95% confidence intervals (CIs) were used to assess the strength of the association. Results Overall, there was no evidence for significant association between MPO−463G>A polymorphism and lung cancer susceptibility (for AA versus GG: OR = 0.91, 95%CI = 0.67–1.24; for GA versus GG: OR = 0.87, 95% CI = 0.78–0.98; for AA/GA versus GG: OR = 0.90, 95% CI = 0.80–1.01; for AA versus GA/GG: OR = 0.96, 95% CI = 0.72–1.28). In the stratified analyses by ethnicity, source of controls and smoking status, we also did not find any significant association between them. Conclusions In summary, this meta-analysis suggests MPO−463G>A polymorphism may not be a risk factor for developing lung cancer. However, further prospective well-designed population-based studies with larger sample size are expected to validate the results.


Introduction
Lung cancer is one of the most common malignancies and the leading cause of cancer-related deaths in the United States and worldwide. It is estimated that in 2011, approximately 221,000 new cases will be diagnosed and 156,900 deaths due to lung cancer will occur in the United States [1]. The exact mechanism of lung carcinogenesis is still unclear. Lung cancer may be a multifactorial disease that resulted from complex interactions between genetic and environmental factors [2]. It has been estimated that cigarette smoking is responsible for 85-90% of lung cancer development [3]. Although smoking can account for the majority of lung cancer, most chronic smokers still do not develop lung cancer. This suggests that lung cancer susceptibility differs among individuals and might have genetic factors which may influence the risk of lung cancer among these who are exposed to tobacco smoke carcinogens.
Myeloperoxidase (MPO) as an endogenous oxidant lysosomal enzyme available in polymorphonuclear neutrophils and monocytes catalyzes the reaction between the chloride ion and hydrogen peroxide and generates hypochlorous acid and other reactive oxygen species (ROS) [4]. The reactive by-products generated by MPO can cause oxidative damage in vivo to biomolecules, such as DNA, protein and lipids, and cause cellular alterations that may lead to carcinogenesis.
The human MPO gene located on chromosome 17q23.1 consists of 12 exons and 11 introns [5] and there are at least 319 different polymorphism sites in the MPO gene (http://www.ncbi. nlm.nih.gov/projects/SNP). The most extensively studied single nucleotide polymorphism (SNP) was 2463G.A polymorphism (rs2333227) located in the promoter region of the MPO gene. Since Austin et al first published the MPO2463G.A polymorphism in 1993 [5], many subsequent studies have consecutively reported the relationship between this polymorphism and different cancer types including esophagus, breast, bladder, brain and lung cancer and so forth. Among them, the relationship between lung cancer risk and the MPO2463G.A polymorphism was the most extensively studied, It has been previously suggested that there was an association between the GG+GA genotype of MPO and a decreased risk of lung cancer [6][7][8][9][10]. However, other studies have failed to confirm such the association [8,[11][12][13][14][15][16][17][18][19][20][21][22][23][24]. The exact relationship between MPO2463G.A polymorphism and susceptibility to lung cancer is inconclusive or conflicting. Up to now, there have been two relevant published meta-analysis studies involving the MPO2463G.A polymorphism and lung cancer risk [25,26], among which one was published in Chinese and the other one was dealt with the meta-analysis on overall cancer susceptibilities. Unfortunately, those two meta-analyses all failed to adopt the most likely appropriate genetic model and lacked subgroups analyses such as smoking status and thus compromised the authentic values of statistical results. In addition, an increasing number of new studies between MPO2463G.A polymorphism and lung cancer risk are available.
So it is necessary and significant to perform a meta-analysis to explore the association between the MPO2463G.A polymorphism and lung cancer risk. Therefore, we performed a metaanalysis on all eligible case-control studies to estimate the overall lung cancer risk of MPO2463G.A polymorphism and to quantify heterogeneity among the individual eligible studies.

Search Strategy
Eligible literatures published before the end of September 2012 were identified by the search of PUBMED, EMBASE, ISI Web of Science databases using combinations of the following keywords: ''Myeloperoxidase'', ''MPO'', ''polymorphism'' or ''variant'' and ''lung cancer'' or ''lung carcinoma'' without restriction on language. All relevant publications were reviewed. Articles in reference lists were also hand-searched for potentially relevant publications.

Inclusion and Exclusion Criteria
Studies included had to meet all the following inclusion criteria: (a) the diagnosis of lung cancer patients were confirmed histologically or pathologically; (b) a case-control study on MPO2463G.A polymorphism and lung cancer; (c) sufficient available data for estimating an odds ratios (ORs) with 95% confidence intervals (CIs). Major reasons for exclusion of studies were as follows: (i) not for lung cancer study, (ii) only case population and (iii) duplicate of previous publication.

Data Extraction
Two investigators (JP Yang, B Wang) extracted the data independently including first author, year of publication, country, ethnicity (Caucasian, Asian and Mixed), source of controls (hospital-based studies, population-based studies), genotyping methods, matching variables, number of genotypes in cases and controls. Discrepancies were adjudicated by the third investigator (YF Zhou) until consensus was achieved on every item. Because two studies [7,11] only provided the information of genotypes as ''GG'' and ''AA/GA'' without data for all three genotypes, we could only calculate the odds ratios (ORs) for the AA/GA versus GG model. We also abstracted the information of smoking status from available studies, and cigarette smoking status was strategically classified as never smokers, light smokers and heavy smokers.

Statistical Analysis
For control group of each study, the allelic frequency was calculated and the observed genotype frequencies of the MPO2463G.A polymorphism were assessed for Hardy-Weinberg equilibrium (HWE) by using the chi-square test. We calculated the strength of the association between MPO2463G.A polymorphism and lung cancer risk by odds ratios (ORs) corresponding to 95% confidence intervals (CIs). The pooled ORs were performed for homozygote model (GA versus GG), heterozygote model (AA versus GG), dominant model (AA/ GA versus GG) and recessive model (AA versus GA/GG) respectively. Stratified analyses were also performed by ethnicity, source of controls and smoking status respectively. Heterogeneity was evaluated with a chi-square-based Q test (P,0.10 was considered significant) [27]. When heterogeneity was present, the random effects model was used to calculate the pooled ORs, whereas the fixed effects model was used [28]. Galbraith plot was used to determine the main sources of the heterogeneity [29,30]. The one-way sensitivity analyses were performed to assess the stability of the results, namely, a single study in the meta-analysis was deleted each time to reect the inuence of the individual data set to the pooled ORs [31]. The publication bias of literatures was assessed using funnel plot and funnel plot asymmetry was assessed by the method of Egger's linear regression test [32][33][34]. The significance of the intercept was determined by the t test suggested by Egger and P,0.05 was considered significant. All statistical analyses were performed with STATA software (version 12.0; STATA Corporation, College Station, TX), and all tests were two sided.

Study Characteristics
For lung cancer susceptibility related to MPO2463G.A polymorphism, Study selection process was shown in Figure 1. 22 case-control studies met the inclusion criteria including7,520 patients with lung cancer and 8,600 controls in this meta-analysis [6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][35][36][37]. The characteristics of included studies and distribution of the frequencies of MPO2463G.A polymorphism on lung cancer were summarized in Table 1 and Table 2 respectively. Overall, there were 9 studies of Caucasians, 6 studies of Asians, 7 of mixed population, and 11 studies of populationbased, 11 studies of hospital-based. Six studies [8,15,16,18,22,23] collected the information on possible confounding factors like smoking status. All cases were pathologically confirmed and Almost controls were mainly matched for age and sex. Most polymorphisms in the control subjects were in Hardy-Weinberg equilibrium.  (Figure 2). In the stratified analysis by ethnicity, source of controls and smoking status, we also did not find any significant association between MPO2463G.A polymorphism and lung cancer risk. MPO 2463G.A polymorphism was a low protective susceptibility gene in lung cancer development in homozygote model (for GA versus GG: OR = 0.87, 95%CI = 0.78-0.98), however, when stratified by ethnicity, source of controls and smoking status, we also did not find any significant association between them (for Caucasian population: OR = 0.86, 95%CI = 0.71-1.05, for Asian population:

Test of Heterogeneity
Significant heterogeneity existed in four genetic models of the MPO2463G.A polymorphism (AA versus GG, GA versus GG, AA/GA versus GG, AA versus GA/GG) ( Table 3). Galbraith plot analyses of all included studies were used to assess the potential sources of heterogeneity. Chan, Liu and London's studies [10,35] were found to be the main contributors of heterogeneity in the AA/GA versus GG model (Figure 3). The significance of pooled ORs with 95%CIs in the AA/GA versus GG model in both overall comparison and subgroup analyses was not influenced by omitting those three studies (Data not shown).

Sensitivity Analysis
Sensitivity analysis was performed through sequentially excluding individual studies. Statistically similar results were obtained after sequentially excluding each study in GA/AA versus GG model ( Figure 4) and the corresponding pooled ORs in the other genetic models were not materially altered (Data not shown), suggesting stability and liability of this meta-analysis.

Publication Bias
The funnel plot and Egger's test did not provide any obvious evidence of publication bias that examined the MPO2463G.A polymorphism and lung cancer risk. The shape of funnel plots did not reveal any evidence of funnel plot asymmetry ( Figure 5), Egger's test further provided statistical evidence of funnel plot symmetry for AA versus GG (P = 0.55), GA versus GG (P = 0.27), AA/GA versus GG (P = 0.56) and for AA versus GA/GG (P = 0.61).

Discussion
As we all know, lung cancer is a complex multifactorial and multistage process, in which both host genetic factors and environmental factors are involved [2]. Individual genetic susceptibility has been suggested to correlate with lung cancer risk. In the current meta-analysis, on the basis of 22 case-control studies providing the information on the MPO2463G.A polymorphism and lung cancer involving 7,520 cases and 8,600 controls, we did not find any significant association between the MPO2463G.A polymorphism and lung cancer risk in any genetic model and also the similar results stratified by ethnicity, source of controls and smoking status respectively. The one-way sensitivity analyses suggested the stability and liability of the results in this meta-analysis. Publication bias was not observed in this study. Our meta-analysis suggests that the MPO 2463G.A polymorphism is not associated with lung cancer development. Oxidative stress occurs when the excessive production of reactive oxygen species (ROS) overwhelms the antioxidant defense system. Increasing evidences suggests variability in these genes involved in oxidative stress may determine the level of oxidative stress in the organism and play a crucial role in carcinogenesis [4,5]. Therefore, it is rational to speculate that certain genetic variants or polymorphisms in the genes involved in oxidative stress may have an impact on cancer risk. MPO is an endogenous oxidant enzyme that generates reactive oxygen species (ROS) that may be involved in carcinogenesis [4,38]. MPO2463G.A polymorphism was located in the promoter region of the MPO gene. The G allele acts as a strong stimulatory protein 1 (SP1) transcription factors binding site, which reacts with SP1 to elevate MPO transcriptional activity [39]. Therefore the guanosine (G) to adenosine (A) nucleotide base substitution is associated with disruption of the SP1 binding site and thus reduces 25 times MPO gene expression and decreases the enzyme levels. So it is biologically plausible that MPO2463G.A polymorphism may modulate the risk of lung cancer. Therefore, many studies have investigated the role of MPO2463G.A polymorphism in the pathogenesis of lung cancer [6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][35][36][37]. However, the results remain conicting rather than conclusive. There are several possible explanations for this discordance, such as small sample size, ethnic background, uncorrected multiple hypothesis testing, and publication bias. Meta-analysis is a statistical procedure for combining the results of several studies to produce a single estimate of the major effect with enhanced precision. So we performed a metaanalysis on 22 eligible case-control studies to estimate the overall lung cancer risk of MPO2463G.A polymorphism, whereas no significant associations were found between them in any genetic model in this meta-analysis. Population stratification is an area of concern that can lead to spurious evidence for the association between a marker and cancer and suggest a possible role for ethnic differences in genetic backgrounds.  Previous studies have found a wide variation in the A allele frequency of the MPO2463G.A polymorphism across different ethnicities. The 2463A allele frequency was 22.8% in European population, but approximately 14.7% in Asian population [40].
When stratified according to ethnicity in this meta-analysis study, no significant associations were found in any of the genetic models in the Caucasian, Asian and mixed population. Although hospitalbased studies may have inherent selection biases, we also did not  find any positive result in the stratified analyses by populationbased and hospital-based studies. These results suggested that the different ethnicities and source of controls did not influence the association between the MPO2463G.A polymorphism and lung cancer risk.
Lung cancer have been characterized as causally related to cigarette smoking [41]. MPO transforms tobacco smoke procarcinogens, such as benzo(a)pyrene and arylamines, into highly carcinogenic intermediates, such as benzo(a)pyrene dio-epoxide. Since MPO2463G.A polymorphism may be associated with weaker transcriptional activity and decreases the enzyme levels, carcinogens contained in cigarette smoking will not be metabolically activated by MPO enzyme, therefore this polymorphism has been suggested to have a protective effect against the development of cancers related to smoking such as lung cancer [42,43]. Whereas in our meta-analysis study, the MPO2463G.A polymorphism was shown to have no statistically significant protective effect in light smokers and heavy smokers. The exact mechanism for this inverse result was not clear. That may be due to the limited number of study subjects related to smoking which may have low power of statistical test, so further large-scale researches between MPO2463G.A polymorphism and risk of lung cancer in smokers are expected to confirm the results.
Several limitations of this meta-analysis should be addressed. First, when interpreting the results of this meta-analysis, heterogeneity was a potential problem and the origins of heterogeneity may include many factors, such as the differences in control characteristics and diverse genotyping methods; Second, the lack of detailed information such as age, sex and lifestyle of the patients in some studies, limited further stratification, and more accurate ORs would be corrected for age, sex and other factors that were associated with lung cancer risk. Nevertheless, our meta-analysis has some advantages. First, the well-designed search and selection method significantly increased the statistical power of this metaanalysis. Second, we could perform a subgroup analysis to address a possible interaction between smoking parameters and MPO2463G.A polymorphism. Third, the results did not show any evidence of publication bias.
In summary, our meta-analysis suggests that the MPO 2463G.A polymorphism may not be a risk factor for lung cancer. However, lung cancer may be a multifactorial disease that resulted from complex interactions between genetic and environmental factors, we could not collect the detailed original data of MPO gene polymorphisms and then we were not able to investigate potential gene-gene, gene-environment interactions. Further prospective researches using adjusted individual data with large sample studies which study the relationship between MPO2463G.A polymorphism and the risk of lung cancer are necessary and expected, which would lead a better, comprehensive understanding of the association between MPO2463G.A polymorphism and lung cancer risk.