Association Between the FAS / FASL Polymorphisms and Gastric Cancer Risk : A Meta-Analysis

OBJECTIVE
FAS/FASL gene promoter polymorphisms have been repeatedly associated with gastric cancer risk, but findings are inconclusive across studies. To address a more precise estimation of the relationship, a meta-analysis was performed.


METHODS
Data were collected from the Pubmed, Medline and EMBASE databases, with the last report up to 1 December, 2011. Crude ORs with 95% CIs were used to assess the strength of the association by (1) the additive, (2) the codominant, (3) the dominant, and (4) the recessive models.


RESULTS
A total of seven studies, including six studies on FAS -1377G>A polymorphism, five studies on FAS -670A>G polymorphism, and six studies on FASL -844T>C polymorphism, were identified in the current meta-analysis. Overall, an association of FAS -1377G>A (AA versus GG: OR = 1.313, 95% CI = 1.045-1.650, Ph = 0.347, I2 = 10.8) and FASL -844T>C (CC versus TT: OR = 1.352, 95% CI = 1.043-1.752, Ph = 0.461, I2 = 0.0) polymorphisms with gastric cancer was found in the codominant model. However, we did not detect any association between gastric cancer and the FAS -670A>G polymorphism. In the subgroup analysis by ethnicity, similar elevated risks were also observed in Asian population for FAS -1377G>A (AA versus GG: OR = 1.309, 95% CI = 1.041- 1.646, Ph = 0.240, I2 = 27.3) and FASL -844T>C (CC versus TT: OR = 1.420, 95% CI = 1.081-1.865, Ph = 0.524, I2 = 0.0) polymorphisms.


CONCLUSIONS
This meta-analysis indicated that FAS -1377G>A and FASL -844T>C polymorphisms might be associated with gastric cancer risk.


Introduction
Gastric cancer (GC) is the fourth most common malignancy and the second leading cause of cancer death worldwide.About one million new cases of GC were estimated to have occurred, followed by the lung, breast and colorectal cancer (Ferlay et al., 2010).However, more than 70% of cases occur in developing countries, and half the world total occurs in Eastern Asia (mainly in China) (Ferlay et al., 2010).Epidemiological studies have suggested several environmental factors may contributed to the development of GC, including cigarette smoking (Yang et al., 2011;Nomura et al., 2012), alcohol consumption (Yang et al., 2006;Duell et al., 2011), pathogenic infections (Yang et al., 2006;Sivachandran et al., 2012) and nutritional deficiency (Yang, 2000).Nevertheless, only a fraction of exposed individuals actually developed GC during their life, suggesting that genetic makeup may confer susceptibility to GC.
Several common low-penetrant genes have been identified as potential GC susceptibility markers.An important one is FAS (also known as TNFSF6, CD95, or APO-1), a cell surface death receptor, which plays an important role in the apoptosis and cancer development (Nagata and Golstein., 1994).By interaction with its natural ligand FASL (also known as CD95L), a member of the tumor necrosis factor superfamily, FAS triggers the death signal cascade contributing to apoptotic cell death (Itoh et al., 1991;Oehm et al., 1992).Aberrant expression of FAS and/or FASL has been detected in many human cancers, including GC (Walboomers et al., 1999;Takahama et al., 2002;Viard-Leveugle et al., 2003).
Over the last two decades, numerous case-control studies have been performed to clarify the relationship between FAS/FASL polymorphisms and GC risk in human (Ikehara et al., 2006;Hsu et al., 2008;Wang et al., 2009;Zhou et al., 2010;Liu et al., 2011;Kupcinskas et al., 2011;Zhang et al., 2011).The most extensively studied polymorphisms are the G to A substitution at position -1377 (-1377G>A, rs2234767) and A to G substitution at position -670 (-670A>G, rs1800682), and the C to T substitution at position -844 (-844C>T, rs763110) in the promoter region of FAS/FASL gene.However, the results of these studies remain conflicting rather than conclusive, partially due to the relative small sample size, different racial and ethnic backgrounds, uncorrected multiple hypothesis testing and publication bias (Zou et al., 2011).To derive a more precise evaluation of the relationship between FAS -1377G>A, -670A>G, and FASL -844T>C polymorphisms and GC susceptibility, we performed a meta-analysis.

Identification of eligible studies
We conducted a comprehensive search on Englishlanguage articles that examined the association of the FAS/FASL gene promoter polymorphisms with GC using Pubmed, Medline and EMBASE database (last report up to 1 December, 2011).Combinations of keywords: ("FAS" or "CD95"), ("FASL" or "CD95L"), ("polymorphism" or "polymorphisms"), "gastric" and ("cancer" or ''carcinoma" or "tumor") were entered as Medical Subject Heading (MeSH) components and as text words.References of identified studies and review articles were checked for other potentially relevant publications.Abstracts or unpublished reports were not considered.If the same patient population was included in several publications, only the study with larger sample size was used in this meta-analysis.For studies including subjects of different ethnic groups, each study should be treated independently.
Eligible studies included in the current meta-analysis should meet the following criterions: (1) it was a casecontrol study; (2) the study was to clarify the association of FAS/FASL polymorphisms with GC; (3) it presented sufficient data to calculate an odds ratio (OR) with 95% confidence interval (CI).We excluded the studies with family members, because their analysis is based on linkage considerations.

Data extraction
Two investigators (Tian J and Pan F) independently extracted the data according to the standard protocol, and the result was reviewed by a third investigator (Ye DQ).Discrepancies were resolved by discussion with our research team.The following information was extracted from each study: the first author's name, publication year, country of origin, racial ancestry, number of genotyped cases and controls, source of control group, genotyping method, control matching method, polymorphisms, studying period and available genotype distributions information.

Meta-analysis methods
Meta-analysis was performed for the polymorphisms that had been investigated in at least three studies.Pooled ORs with 95% CIs were used to assess the strength of association between the FAS/FASL polymorphisms and susceptibility to GC.We evaluated the risk of (1) additive model (minor allele versus major allele); (2) codominant model (heterozygous versus common homozygous carriers and rare homozygous versus common homozygous carriers); (3) dominant model (rare allele carriers versus common homozygous carriers); (4) recessive model (rare homozygous carriers versus common allele carriers).The between-study heterogeneity was assessed using the Chi-square test-based Q-statistic (Cochran, 1954).If a significant Q-statistic (P < 0.1) was observed, indicating heterogeneity across studies, the random-effects model was used (DerSimonian and Laird, 1986).Otherwise, the fixed-effect model would be explored (Mantel and Haenszel, 1959).The random-effects model assumes different studies show substantial diversity and assesses both within-study sampling error and between-study variation (DerSimonian and Laird, 1986).The fixed-effect model assumes that all of the studies are estimating the same underlying effect and considers only within-study variance (Mantel and Haenszel, 1959).We also quantified the effect of heterogeneity using I 2 = 100%×(Q-df)/Q (Higgins and Thompson, 2002), which ranges between 0 and 100%, and measures the degree of inconsistency in the studies by calculating what proportion of the total variation across studies attributed to heterogeneity rather than chance (Higgins et al., 2003).The overall estimate of risk was obtained by DerSimonian and Laird method in a random-effects model or Mantel-Haenszel method in a fixed-effects model in the presence (P ≤ 0.1 or I 2 > 50%) or absence (P > 0.1 or I 2 ≤ 50%) of heterogeneity, respectively (Mantel and Haenszel, 1959;DerSimonian and Laird, 1986).Pooled OR with 95% CI was performed weighting individual OR by the inverse of their variance.The significance of the pooled OR was determined by the Z-test.
A chi-square test was used to estimate the Hardy-Weinberg equilibrium (HWE) among the control individuals to compare the observed genotype frequencies with the expected ones.The power analysis of each study was done using the statistical program G *Power 3.1 at the level 0.05 level of significance, assuming an OR of 1.5 (small effect size) (Faul et al., 2009).

Evaluation of publication bias
We estimated the potential publication bias by the funnel plot, in which the standard error of log (OR) of each study was plotted against its log (OR).If there was publication bias, the funnel plot would be asymmetric.Funnel plot asymmetry was further determined by the method of Egger's linear regression test (Egger et al., 1997), which measures funnel plot asymmetry on the natural logarithm scale of the OR.The significance of the intercept was determined by the t-test, and P < 0.05 was considered representative of statistically significant publication bias.
All the statistical analyses were conducted by STATA version 7.0 (Stata Corporation, College Station, TX, USA).A P-value less than or equal to 0.05 was considered statistically significant.

Studies included in the meta-analysis
A total of seven studies, six studies for FAS -1377G>A and FASL -844T>C polymorphisms, respectively, and five studies for FAS -670A>G polymorphism, met the inclusion criteria (Ikehara et al., 2006;Hsu et al., 2008;Wang et al., 2009;Zhou et al., 2010;Kupcinskas et al., 2011;Liu et al., 2011;Zhang et al., 2011).The characteristics of each article are listed in Table 1.Seven separate studies consisted of six Asian and one Caucasian.Of these articles, six studies were hospital-based, and one study was population-based.We calculated the expected power of each study to demonstrate an association between FAS/FASL polymorphisms and GC (Table 1).The results of HWE test for the genotypes distributions in control population are shown in Table 2.All the eligible studies were consistent in HWE.

Meta-analysis
A summary of the meta-analysis for the FAS/FASL promoter polymorphisms and GC risk is given in Table 3.

Evaluation of FAS -1377G>A polymorphism and association with GC
The association between FAS -1377G>A polymorphism and GC was investigated in six separate studies including 1513 cases and 2007 controls.No significant heterogeneity was observed and the original data were combined by means of the fixed-effects models.An association of FAS -1377G>A polymorphism with GC was found in the contrast of AA versus GG (OR = 1.313, 95% CI = 1.045-1.650,Ph = 0.347, I 2 = 10.8) when all studies were pooled into the meta-analysis.Owing to the limited literature in Caucasian population and population-based controls, subgroup stratification was only performed in Asian population and hospital-based studies.Similar association was also found in Asians (OR = 1.309, 95% CI = 1.041-1.646,Ph = 0.240, I 2 = 27.3) and hospital-based studies (OR = 1.333, 95% CI = 1.033-1.722,Ph = 0.237, I 2 = 27.6).

Evaluation of FASL -844T>C polymorphism and association with GC
We found six separate studies (1513 cases and 2007 controls) investigating the association between FASL -844T>C polymorphism and GC risk.The Q test of heterogeneity was not significant and we conducted analyses using the fixed-effects models, except in the comparison of C versus T.An association was found in the overall population when examining the contrast of CC versus TT (OR = 1.352, 95% CI = 1.043-1.752,Ph = 0.461, I 2 = 0.0).Meanwhile, we performed group-specific meta-analysis in Asian population and hospital-based DOI:http://dx.doi.org/10.7314/APJCP.2012.13.3.945Association Between the FAS/FASL Polymorphisms and Gastric Cancer Risk: A Meta-analysis studies.Similarly, elevated risks were observed among Asians (OR = 1.420, 95% CI = 1.081-1.865,Ph = 0.524, I 2 = 0.0) and groups with hospital-based controls for CC versus TT (OR = 1.353,CI = 1.013-1.807,Ph = 0.326, I 2 = 13.8).

Publication bias
The shapes of the funnel plots revealed no obvious asymmetry (figures not shown).Then, the Egger's test was used to provide statistical evidence of funnel plot symmetry.Also, the results still did not suggest any evidence of publication bias (Table 4).

Discussion
Apoptosis is one of the most important regulatory mechanisms to all multicellular organisms for normal development and tissue homeostasis (Reed, 2000).Inappropriate regulation of apoptosis contributes to a number of human disorders, including GC (Thompson, 1995;Hajra and Liu, 2004).There were two main apoptotic pathways in mammalian cells: the extrinsic or receptormediated pathway and the intrinsic or mitochondrial pathway (Nicholson and Thornberry, 1997;Ashkenazi and Dixit, 1999;Budihardjo et al., 1999).FAS is a cell surface receptor that belongs to the tumor necrosis factor receptor family.By interaction with its natural ligand FASL, FAS initiates the extrinsic apoptotic pathway (Itoh et al., 1991;Oehm et al., 1992;Suda et al., 1993).Accumulating evidence showed that aberrant expression of FAS and FASL in many human cancers, including GC (Walboomers et al., 1999;Takahama et al., 2002;Viard-Leveugle et al., 2003;Gryko et al., 2011).It has been proposed that down-regulation of FAS may protect tumor cells from elimination by anti-tumor immune responses, whereas upregulation of FASL may increase the ability of tumor cells to counterattack the immune system by inducing apoptosis of FAS-sensitive lymphocytes (Griffith et al., 1995;Strand et al., 1996;Reichmann, 2002).Therefore, it is reasonable to speculate that FAS/FASL system may play a crucial role in the pathogenesis of GC.In recent years, genetic variants of the FAS/FASL in GC have drawn increasing attention.Growing number of studies have suggested that the -1377G>A and -670A>G polymorphisms in the promoter region of FAS gene, and the -844T>C polymorphism in the promoter region of FASL were emerging as susceptibility loci for GC.However, the results were inconclusive.To better understand the relationship between FAS/FASL polymorphisms (FAS -1377G>A, -670A>G and FASL -844T>C) and GC risk, a meta-analysis was performed.
Overall, our results indicated that the variant genotypes of the FAS -1377G>A and FASL -844T>C polymorphisms but not the FAS -670A>G polymorphism were associated with susceptibility to GC (FAS -1377G>A: AA vs. GG: OR = 1.313, 95% CI = 1.045-1.650,Ph = 0.347, I 2 = 10.8;FASL -844T>C: CC vs. TT: OR = 1.352, 95% CI = 1.043-1.752,Ph = 0.461, I 2 = 0.0).This finding is biologically plausible.It has been proven that as compared with the -1377G allele, the -1377A allele had a greatly reduced ability to bind transcription factor stimulatory protein 1, whereas the -670A and G alleles had similar ability to bind transcription factor signal transducers and activators of transcription 1 (Sibley et al., 2003).As an important transcriptional activator, if the binding ability of stimulatory protein 1 to the FAS -1377A allele is reduced, decreased expression of FAS in cells carrying the FAS -1377AA genotype was expected (Huang et al., 1997;Sibley et al., 2003).It has been shown that the FASL -844T>C polymorphism has a substantial impact on promoter activity of the FASL gene in an in vitro assay system because of its location in a binding motif for transcription factor CAAT/ enhancer-binding protein β (Wu et al., 2003).Moreover, this variation strongly affected the FASL expression on ex vivo-stimulated T cells (Sun et al., 2005).Activation-induced cell death (AICD) of T lymphocytes may help malignant cells to escape from killing by natural killing cells (Chappell and Restifo, 1998;Green et al., 2003).It has been proposed that the FASL -844C allele had a higher expression on T cells and was associated with an enhanced rate of AICD of T cells, which may result in less powerful immune surveillance and increase the susceptibility to cancer compared with the -844T allele (Sun et al., 2005).
Conspicuous geographic variation exists in the incidence of GC between regions.The highest incidence is in northeast Asia, intermediate incidences occur in Europe and South America, and North America, Africa, south Asia and Oceania are low incidence regions (Hartgrink et al., 2009).Population differences may enlighten some genetic risk factors that are specific towards certain ethnic groups, which may help elucidate the ethnic differences in terms of prevalence and severity.To explore whether the FAS/FASL polymorphisms are associated with GC risk in different genetic backgrounds, subgroup analysis based on ethnicity was performed.We found an association of FAS -1377 G>A and FASL -844T>C polymorphisms with GC among Asians (FAS -1377G>A: AA vs. GG: OR = 1.309, 95% CI = 1.041-1.646,Ph = 0.240, I 2 = 27.3;FASL -844T>C: CC vs. TT: OR = 1.420, 95% CI = 1.081-1.865,Ph = 0.524, I 2 = 0.0).Similar association was not replicated in Caucasian population, suggesting a possible role of ethnic differences and the environment they lived in (Hirschhorn et al., 2002).Other factors such as selection bias and different matching criteria may also play a role.Considering only one study carried out in Caucasian population, the result might be not reliable.
Some limitations in this meta-analysis should be acknowledged.Firstly, most of eligible studies involved in the current meta-analysis were hospital-based case-control studies, which inevitably suffer selection bias (Knottnerus., 1987).However, each study was in HWE, suggesting the controls could well represent the general population.Secondly, there might be a potential English language bias in the current study, because this meta-analysis only contained the English literature.It was possible that there were differences between English language literature and other language literature.Thirdly, in the subgroup analysis based on ethnicity, only one study containing 271 cases and 271 controls was performed in Caucasian population, there may not be enough statistical power to obtain the real relationship.Thus, our result of subgroup metaanalysis should be interpreted with caution, and further studies with larger sample size are still needed, especially in Caucasians.Fourthly, this meta-analysis was based on unadjusted estimates, while a more precise analysis could be performed if individual data was available, which would allow for an adjustment estimate by other co-variants, including age, sex, and environmental factors.
meta-analysis remains a retrospective research, which is subject to the methodological deficiencies of the included studies.To minimize the likelihood of bias, we developed a detailed protocol before initiating the study, performed a meticulous search for eligible studies and used explicit methods for data extraction and statistical analysis.
In summary, our meta-analysis suggests that there may be an association of FAS -1377 G>A and FASL -844 T>C polymorphisms with GC.To reach a definitive conclusion, further gene-gene and gene-environment interaction studies based on larger sample size are still needed.