Meta-Analysis of the Association between the rs 8034191 Polymorphism in AGPHD 1 and Lung Cancer Risk

Lung cancer is the leading cause of cancer-related death throughout the worldwith an estimated 1.3 million new cases diagnosed annually (Shibuya et al.,2002; Herbst et al., 2008). In many countries,the morbidity and mortality of lung cancer have increased rapidly in recent years (Bhat et al., 2013; Shukla et al., 2013; Yilmaz et al., 2014). Well-known risk factors for lung cancer include cigarette smoking and exposure to ionizing radiation. Although over 80% of lung cancer cases are related to the use of tobacco (Parkin et al., 1994), only a small percentage of smokers (<20%) develop this disease. A ccumulating evidence suggests that genetic factors may contribute to variation in susceptibility to lung cancer. It is widely accepted that lung cancer is a complex multifactorial disease that is attributed to the interaction of genetic factors with environmental factors (Amos et al., 2008; Heller et al., 2010; ).Despite intensive efforts devoted to investigating the genetic factors associated with lung cancer, the genes and genetic variants that drive the development of lung cancer remain unclear.


Introduction
Lung cancer is the leading cause of cancer-related death throughout the worldwith an estimated 1.3 million new cases diagnosed annually (Shibuya et al.,2002;Herbst et al., 2008).In many countries,the morbidity and mortality of lung cancer have increased rapidly in recent years (Bhat et al., 2013;Shukla et al., 2013;Yilmaz et al., 2014).Well-known risk factors for lung cancer include cigarette smoking and exposure to ionizing radiation.Although over 80% of lung cancer cases are related to the use of tobacco (Parkin et al., 1994), only a small percentage of smokers (<20%) develop this disease.A ccumulating evidence suggests that genetic factors may contribute to variation in susceptibility to lung cancer.It is widely accepted that lung cancer is a complex multifactorial disease that is attributed to the interaction of genetic factors with environmental factors (Amos et al., 2008;Heller et al., 2010;).Despite intensive efforts devoted to investigating the genetic factors associated with lung cancer, the genes and genetic variants that drive the development of lung cancer remain unclear.

Publication search andinclusion/exclusion criteria
In August 2013, we searched PubMed, Google Scholar, EMBASE, and the ChinaNational Knowledge Infrastructure using the following search terms: AGPHD1, rs8034191, lung cancer, gene, genotype, mutation, and polymorphism.Articles identifiedin the primary literaturemet our initial criteria for inclusion in the metaanalysis if they were published in English, focused on humans, and were free of obvious overlap with other studies.
From the publications identified above, two investigators independently selected articles containing information on the association between AGPHD1 and lung cancer morbidity and checked the corresponding reference lists.If multiple studies were published on the same population or subpopulation, only the most recent or informative study was included in the meta-analysis.
Articles were included in this meta-analysis if they 1) examinedthe hypothesis that rs8034191is associated withlung cancer risk, 2) followed a nested case-control, case-control,or cross-sectional study design, and 3) provided estimates of ORs and corresponding 95% CIs or sufficientinformation on genotype/allele counts between cases and controlsto calculatethe ORsand 95%CIs.Articles were excluded if they included non-case-control studies, a control population containing patients with malignant tumor, orwere redundant with other published studies.A flow chart outlining the selection process for inclusion in the meta-analysis is shown in Figure 1.

Data extraction
The following information was extracted from each study: the first author's name, the year of publication, the countryin which the study was performed, ethnicities of subjects, andthe number of cases and controlswith the TT, TC, and CC rs8034191 genotypes.Two investigators independently extracted the data from all eligible publications, and any inconsistencies were resolved by discussion.
All statistical analyses were performed using STATA software (version 11.0; Stata Corporation, College Station, TX).Two-sided p-values less than 0.05 were considered statistically significant.We calculated the allelic frequenciesfor the case and control groups in each studyand assessed them for Hardy-Weinberg equilibrium (HWE) using achi-square test (Egger et al.,1997).The OR and 95% CI valueswere determinedto assess the strength of the association between each rs8034191 polymorphism and lung cancer risk.For each study, the OR and 95% CI were assessed in adominant model, a recessive model, and an additive model.Subgroup analyses were performed based on the source of the controls and the ethnicity of the study participants.The chi-squared based Q-statistic was calculated to test for heterogeneity among the studies.If the studies were found to be heterogeneous (p<0.05) the pooled ORs were analyzed using a random-effects model (Higgins et al., 2002); otherwise, a fixed-effects model was used (Egger et al.,1997).The I 2 statistic was then used to quantitatively estimate heterogeneity, with I 2 less than 25%, between 25% and 75%, and greater than 75% representing low, moderate, and high degrees of inconsistency, respectively (Higgins et al., 2003;Hemminki et al., 2006).The significance of the combined OR was determinedusing a Z test (p<0.05was considered statistically significant).Cumulative meta-analyses were performed on all eligible cancer studies according to case sample size.Additionally, sensitivity of the metaanalysis was evaluated throughthe sequential removal of each study.
Finally, we produced aBegg's funnel plot and performed an Egger's test tostatistically assess publication bias.The Egger's test was based on the linear regression of the standard normal deviate against the precision of the standard normal deviate and was used to test the symmetry of the Begg's funnel plot.p<0.05 was considereda significant publication bias (Higgins et al., 2002).

Eligible studiesand quality assessment
Through the procedure outlined in Figure 1 we identified ninearticles that met the inclusion criteria.These articles covered13 case-control studies including 16,858 cases and 19,576 controls.The characteristics of the studies are listed in Table 1.The sestudies focused solely on lung cancer and represented multiple ethnic populations.The alleles at rs8034191were in HWE in all of the studies.After evaluating these studies, all 13 were deemed to be of sufficient quality to be included in our analysis.
There was no significant heterogeneity in the rs8034191 variant genemodel.We stratified the data by dividing the participants into two subgroups based on ethnicity: Asian and Caucasian.The pooled Ors for the recessive model were 1.887 (95% CI 0.489-7.274)for the Asian subgroup and1.412(95% CI 1.323-1.508)for the Caucasian subgroup (Table 2).

Sensitivity of the meta-analysis
The pooled OR values were not qualitatively changed by the elimination of any individual study (data not shown), indicating that the final results of the metaanalysis were relatively stable and reliable.Likewise, the results of the meta-analysis were not influenced by departure of the allele frequency from HWE .

Publication bias
A major concern for any meta-analysis is the potential introduction of a publication bias based on the inclusion/ exclusion criteria used to select studies for the analysis.To investigate whether publication bias was present in this study, a Begg's funnel plot was constructed (Figure 5).The shape of the Begg's funnel plot was relatively symmetric, indicating there was no obvious publication bias.An Egger's test based on this plot revealed no statistically significant asymmetry in the funnel plot for any of the genetic models (p=0.933).

Discussion
We performeda comprehensive meta-analysis to evaluate the associationof a common polymorphismon 15q25.1 with the risk of lung cancer.By performing subgroup analyses, we identified ethnicity as a potential source of inconsistency between studies.This is not surprising, as it is well established that genetic heterogeneity isinevitable in disease identification strategies (Higgins et al., 2003).Specifically, the overall results of this study demonstrated that the rs8034191-T allele of the AGPHD1 gene might be a risk factor for the development of lung cancer in Caucasians, but not in Asians (Table 2).We also noticed remarkable differences inthe rs8034191-C allele between Caucasians and Asians, making it very difficult to detect weak associations in Asians unless examininga very large population.This suggests that different genetic backgrounds may differentially affect this allele or that different populations may have different linkage disequilibrium patterns; for example, the studied polymorphisms may be in linkage with another causalvariant in one ethnic population but not in another (Hemminki et al., 2006).Considering the divergent genetic backgrounds, it is necessary to construct a database of polymorphisms related to lung cancer ineach ethnic/racial group.
To the best of our knowledge, the present study is the only meta-analysis to date investigating the association of the rs8034191 polymorphism with lung cancer susceptibility.Although potential sources of heterogeneity cannot be easily eliminated, the strengths of this study include the relatively large sample size, the lack of deviation from Hardy-Weinberg equilibrium, and the highquality of the studies involved.However, this study should be interpreted with several technical limitations in mind.First, most of the studies in this meta-analysis were case-control studies, which are susceptible to selection bias by including only non-fatalcases.Second, because only studies in the English language were considered, a publication bias may have been introduced.To address this, we performed a funnel plot and an accompanying Egger's test, which did not reveal an obvious bias.Moreover, any asymmetry in the funnel plot, either through visually interpretation or statistically testing, may result from a fundamental difference between small and large studies that arises frominherent inter-study heterogeneity There is no perfect method to test for publication bias, and the validity of the funnel plot and Egger's test have been challenged (Yu et al., 2010).Thus, we cannot completely rule out the low probability that relevant studies (for example small negative studies) are missing from the plot although the trim and fill method suggested that no missing studies were required to make the funnel plot symmetrical for either polymorphism.Third, the single locus-based nature of this meta-analysis precluded the possibility of investigating gene-geneand gene-environment interactions, as well as haplotype-based effects.In particular, further studies should investigate other markers adjacent to 15q25.1 to clarify whether the observed association is causal or due to linkage disequilibrium.It is likely that the rs8034191SNP alone makes aminor contribution to risk prediction in lung cancer patients,and further studies are needed to determine whether multiple polymorphisms integrated with other riskfactors will enhance the predictive capabilities.Additional studies are necessary to fully understand the relationship between the rs8034191 SNP and lung cancer susceptibility.
In conclusion, we have expanded previous studies by providing evidence thatthe rs8034191-T allele of the AGPHD1 gene mightbe a risk factor for the development of lung cancer inCaucasians, but not in Asians.Functional studies of the association between this polymorphism and cancer risk are warranted.

Figure 2 .
Figure 2. Overall Meta-analysis of the AGPHD1 rs8034191 Polymorphism and Lung Cancer Risk in the Dominant Genetic Mode

Figure
Figure 3. Overall Meta-analysis of the AGPHD1 rs8034191 Polymorphism and Lung Cancer Risk in the Recessive Genetic Mode

Figure 4 .
Figure 4. Overall Meta-analysis of the AGPHD1 rs8034191 Polymorphism And Lung Cancer risk in the Additive Genetic Mode